Domain Generalization (DG) has been recently explored to enhance the generalizability of Point Cloud Classification (PCC) models toward unseen domains. Prior works are based on convolutional networks, Transformer or Mamba architectures, either suffering from limited receptive fields or high computational cost, or insufficient long-range dependency modeling. RWKV, as an emerging architecture, possesses superior linear complexity, global receptive fields, and long-range dependency. In this paper, we present the first work that studies the generalizability of RWKV models in DG PCC. We find that directly applying RWKV to DG PCC encounters two significant challenges: RWKV's fixed direction token shift methods, like Q-Shift, introduce spatial distortions when applied to unstructured point clouds, weakening local geometric modeling and reducing robustness. In addition, the Bi-WKV attention in RWKV amplifies slight cross-domain differences in key distributions through exponential weighting, leading to attention shifts and degraded generalization. To this end, we propose PointDGRWKV, the first RWKV-based framework tailored for DG PCC. It introduces two core modules to enhance spatial modeling and cross-domain robustness, while maintaining RWKV's linear efficiency. In particular, we present Adaptive Geometric Token Shift to model local neighborhood structures to improve geometric context awareness. In addition, Cross-Domain key feature Distribution Alignment is designed to mitigate attention drift by aligning key feature distributions across domains. Extensive experiments on multiple benchmarks demonstrate that PointDGRWKV achieves state-of-the-art performance on DG PCC.

Fangda Ye, Kuicai Dong, Xie Zhifei, Yuxin Hu, Yihang Yin, Shurui Huang, Shikai Dong, Chen Zhang, Jianzhu Bao, Shuicheng Yan. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.

Guibin Zhang, Haiyang Yu, Kaiming Yang, Bingli Wu, Fei Huang, Yongbin Li, Shuicheng Yan. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.

Mixture-of-Experts (MoE) has emerged as an effective and efficient scaling mechanism for large language models (LLMs) and vision-language models (VLMs). By expanding a single feed-forward network into multiple expert branches, MoE increases model capacity while maintaining efficiency through sparse activation. However, despite this sparsity, the need to preload all experts into memory and activate multiple experts per input introduces significant computational and memory overhead. The expert module becomes the dominant contributor to model size and inference cost, posing a major challenge for deployment. To address this, we propose MC# (Mixture-Compressor-sharp), a unified framework that combines static quantization and dynamic expert pruning by leveraging the significance of both experts and tokens to achieve aggressive compression of MoE-LLMs/VLMs. To reduce storage and loading overhead, we introduce Pre-Loading Mixed-Precision Quantization (PMQ), which formulates adaptive bit allocation as a linear programming problem. The objective function jointly considers expert importance and quantization error, producing a Pareto-optimal trade-off between model size and performance. To reduce runtime computation, we further introduce Online Top-any Pruning (OTP), which models expert activation per token as a learnable distribution via Gumbel-Softmax sampling. During inference, OTP dynamically selects a subset of experts for each token, allowing fine-grained control over activation. By combining PMQ's static bit-width optimization with OTP's dynamic routing, MC# achieves extreme compression with minimal accuracy degradation. On DeepSeek-VL2, MC# achieves a 6.2 × weight reduction at an average of 2.57 bits, with only a 1.7% drop across five multimodal benchmarks compared to the 16-bit baseline. Moreover, OTP further reduces expert activation by 20% with less than 1% performance loss, demonstrating strong potential for efficient deployment of MoE-based models.

This paper presents UniVST, a unified framework for localized video style transfer based on diffusion models. It operates without the need for training, offering a distinct advantage over existing diffusion methods that transfer style across entire videos. The endeavors of this paper comprise: (1) A point-matching mask propagation strategy that leverages the feature maps from the DDIM inversion. This streamlines the model's architecture by obviating the need for tracking models. (2) A training-free AdaIN-guided localized video stylization mechanism that operates at both the latent and attention levels. This balances content fidelity and style richness, mitigating the loss of localized details commonly associated with direct video stylization. (3) A sliding-window consistent smoothing scheme that harnesses optical flow within the pixel representation and refines predicted noise to update the latent space. This significantly enhances temporal consistency and diminishes artifacts in stylized video. Our proposed UniVST has been validated to be superior to existing methods in quantitative and qualitative metrics. It adeptly addresses the challenges of preserving the primary object's style while ensuring temporal consistency and detail preservation.

Generating intermediate video content of varying lengths based on given first and last frames, along with text prompt information, offers significant research and application potential. However, traditional frame interpolation tasks primarily focus on scenarios with a small number of frames, no text control, and minimal differences between the first and last frames. Recent community developers have utilized large video models represented by Wan to endow frame-to-frame capabilities. However, these models can only generate a fixed number of frames and often fail to produce satisfactory results for certain frame lengths, while this setting lacks a clear official definition and a well-established benchmark. In this paper, we first propose a new practical Semantic Frame Interpolation (SFI) task from the perspective of academic definition, which covers the above two settings and supports inference at multiple frame rates. To achieve this goal, we propose a novel SemFi model building upon Wan2.1, which incorporates a Mixture-of-LoRA module to ensure the generation of high-consistency content that aligns with control conditions across various frame length limitations. Furthermore, we propose SFI-300K, the first general-purpose dataset and benchmark specifically designed for SFI. To support this, we collect and process data from the perspective of SFI, carefully designing evaluation metrics and methods to evaluate the performance of the model in multiple dimensions, including images and videos, and various aspects, including consistency and diversity. Through extensive experiments on SFI-300K, we demonstrate that our method is particularly well-suited to meet the requirements of the SFI task.

We introduce DiffusionTrend for virtual fashion try-on, which forgoes the need for retraining diffusion models. Using advanced diffusion models, DiffusionTrend harnesses latent information rich in prior information to capture the nuances of garment details. Throughout the diffusion denoising process, these details are seamlessly integrated into the model image generation, expertly directed by a precise garment mask crafted by a lightweight and compact CNN. Although our DiffusionTrend model initially demonstrates suboptimal metric performance, our exploratory approach offers some important advantages: (1) It circumvents resource-intensive retraining of diffusion models on large datasets. (2) It eliminates the necessity for various complex and user-unfriendly model inputs. (3) It delivers a visually compelling try-on experience, underscoring the potential of training-free diffusion model. This initial foray into the application of untrained diffusion models in virtual try-on technology potentially paves the way for further exploration and refinement in this industrially and academically valuable field.

Recent advancements in multimodal large language models (LLMs) have demonstrated significant potential across various domains, particularly in concept reasoning. However, their applications in understanding 3D environments remain limited, primarily offering textual or numerical outputs without generating dense, informative segmentation masks. This paper introduces Reason3D, a novel LLM designed for comprehensive 3D understanding. Reason3D processes point cloud data and text prompts to produce textual responses and segmentation masks, enabling advanced tasks such as 3D reasoning segmentation, hierarchical searching, express referring, and question answering with detailed mask outputs. We propose a hierarchical mask decoder that employs a coarse-to-fine approach to segment objects within expansive scenes. It begins with a coarse location estimation, followed by object mask estimation, using two unique tokens predicted by LLMs based on the textual query. Experimental results on large-scale ScanNet and Matterport3D datasets validate the effectiveness of our Reason3D across various tasks.

In-context segmentation has drawn increasing attention with the advent of vision foundation models. Its goal is to segment objects using given reference images. Most existing approaches adopt metric learning or masked image modeling to build the correlation between visual prompts and input image queries. This work approaches the problem from a fresh perspective - unlocking the capability of the latent diffusion model (LDM) for in-context segmentation and investigating different design choices. Specifically, we examine the problem from three angles: instruction extraction, output alignment, and meta-architectures. We design a two-stage masking strategy to prevent interfering information from leaking into the instructions. In addition, we propose an augmented pseudo-masking target to ensure the model predicts without forgetting the original images. Moreover, we build a new and fair in-context segmentation benchmark that covers both image and video datasets. Experiments validate the effectiveness of our approach, demonstrating comparable or even stronger results than previous specialist or visual foundation models. We hope our work inspires others to rethink the unification of segmentation and generation.

Recent advancements in multimodal large language models (MLLMs) have shown unprecedented capabilities in advancing various vision-language tasks. However, MLLMs face significant challenges with hallucinations, and misleading outputs that do not align with the input data. While existing efforts are paid to combat MLLM hallucinations, several pivotal challenges are still unsolved. First, while current approaches aggressively focus on addressing errors at the perception level, another important type at the cognition level requiring factual commonsense can be overlooked. In addition, existing methods might fall short in finding a more effective way to represent visual input, which is yet a key bottleneck that triggers visual hallucinations. Moreover, MLLMs can frequently be misled by faulty textual inputs and cause hallucinations, while unfortunately, this type of issue has long been overlooked by existing studies. Inspired by human intuition in handling hallucinations, this paper introduces a novel bottom-up reasoning framework. Our framework systematically addresses potential issues in both visual and textual inputs by verifying and integrating perception-level information with cognition-level commonsense knowledge, ensuring more reliable outputs. Extensive experiments demonstrate significant improvements in multiple hallucination benchmarks after integrating MLLMs with the proposed framework. In-depth analyses reveal the great potential of our methods in addressing perception- and cognition-level hallucinations.

Domain Generalization (DG) has been recently explored to improve the generalizability of point cloud classification (PCC) models toward unseen domains. However, they often suffer from limited receptive fields or quadratic complexity due to the use of convolution neural networks or vision Transformers. In this paper, we present the first work that studies the generalizability of state space models (SSMs) in DG PCC and find that directly applying SSMs into DG PCC will encounter several challenges: the inherent topology of the point cloud tends to be disrupted and leads to noise accumulation during the serialization stage. Besides, the lack of designs in domain-agnostic feature learning and data scanning will introduce unanticipated domain-specific information into the 3D sequence data. To this end, we propose a novel framework, PointDGMamba, that excels in strong generalizability toward unseen domains and has the advantages of global receptive fields and efficient linear complexity. PointDGMamba consists of three innovative components: Masked Sequence Denoising (MSD), Sequence-wise Cross-domain Feature Aggregation (SCFA), and Dual-level Domain Scanning (DDS). In particular, MSD selectively masks out the noised point tokens of the point cloud sequences, SCFA introduces cross-domain but same-class point cloud features to encourage the model to learn how to extract more generalized features. DDS includes intra-domain scanning and cross-domain scanning to facilitate information exchange between features. In addition, we propose a new and more challenging benchmark PointDG-3to1 for multi-domain generalization. Extensive experiments demonstrate the effectiveness and state-of-the-art performance of PointDGMamba.

Recently, state space models have exhibited strong global modeling capabilities and linear computational complexity in contrast to transformers. This research focuses on applying such architecture to more efficiently and effectively model point cloud data globally with linear computational complexity. In particular, for the first time, we demonstrate that Mamba-based point cloud methods can outperform previous methods based on transformer or multi-layer perceptrons (MLPs). To enable Mamba to process 3-D point cloud data more effectively, we propose a novel Consistent Traverse Serialization method to convert point clouds into 1-D point sequences while ensuring that neighboring points in the sequence are also spatially adjacent. Consistent Traverse Serialization yields six variants by permuting the order of x, y, and z coordinates, and the synergistic use of these variants aids Mamba in comprehensively observing point cloud data. Furthermore, to assist Mamba in handling point sequences with different orders more effectively, we introduce point prompts to inform Mamba of the sequence’s arrangement rules. Finally, we propose positional encoding based on spatial coordinate mapping to inject positional information into point cloud sequences more effectively. Point Cloud Mamba surpasses the state-of-the-art (SOTA) point-based method PointNeXt and achieves new SOTA performance on the ScanObjectNN, ModelNet40, ShapeNetPart, and S3DIS datasets. It is worth mentioning that when using a more powerful local feature extraction module, our PCM achieves 79.6 mIoU on S3DIS, significantly surpassing the previous SOTA models, DeLA and PTv3, by 5.5 mIoU and 4.9 mIoU, respectively.

In this work, we investigate how to sparsify a pre-trained dense large language model into a mixture-of-experts (MoE) architecture for faster inference.Our approach applies mask matrix to the activations for each expert, constrained by L 0 regularization to minimize the number of activated parameters.To ensure minimal performance loss under this constraint, we initialize the model with all parameters active and progressively sparsify it during training.This approach proves more efficient than oneshot sparsification techniques, which typically require significant resources for performance recovery.Moreover, our approach automatically identifies shared, token-specific, and inactive experts, allowing for more efficient allocation of computational resources.Through extensive experiments, we achieve up to 97% performance retention on downstream tasks with only 50% of the feed-forward parameters activated in dense models.Beyond improving inference efficiency, this strategy of sharing computational units among experts provides a principled foundation for building more scalable and generalizable MoE architectures, paving the way for future expert-based model designs.Our code is available at https:// github.com/lpyhdzx/Mixture-of-Masks.

Recent advancements in multimodal reasoning overlook the audio modality.We introduce Audio-Reasoner, a large-scale audio language model for deep reasoning.We meticulously curated a large-scale and diverse multi-task audio dataset with simple annotations.Then, we leverage closed-source models to conduct secondary labeling, QA generation, along with structured COT process.These datasets together form a high-quality reasoning dataset with 1.2 million reasoning-rich samples, which we name CoTA.Following inference scaling principles, we train Audio-Reasoner on CoTA, enabling it to achieve great logical capabilities in audio reasoning.Experiments show state-ofthe-art performance across key benchmarks, including MMAU-mini (+25.42%),AIR-Bench chat/foundation (+14.57%/+10.13%),and MELD (+8.01%).Our findings stress the core of structured CoT training in advancing audio reasoning.

4D Gaussian Splatting (4DGS) has recently emerged as a promising technique for capturing complex dynamic 3D scenes with high fidelity. It utilizes a 4D Gaussian representation and a GPU-friendly rasterizer, enabling rapid rendering speeds. Despite its advantages, 4DGS faces significant challenges, notably the requirement of millions of 4D Gaussians, each with extensive associated attributes, leading to substantial memory and storage cost. This paper introduces a memory-efficient framework for 4DGS. We streamline the color attribute by decomposing it into a per-Gaussian direct color component with only 3 parameters and a shared lightweight alternating current color predictor. This approach eliminates the need for spherical harmonics coefficients, which typically involve up to 144 parameters in classic 4DGS, thereby creating a memory-efficient 4D Gaussian representation. Furthermore, we introduce an entropy-constrained Gaussian deformation technique that uses a deformation field to expand the action range of each Gaussian and integrates an opacity-based entropy loss to limit the number of Gaussians, thus forcing our model to use as few Gaussians as possible to fit a dynamic scene well. With simple half-precision storage and zip compression, our framework achieves a storage reduction by approximately 190$\times$ and 125$\times$ on the Technicolor and Neural 3D Video datasets, respectively, compared to the original 4DGS. Meanwhile, it maintains comparable rendering speeds and scene representation quality, setting a new standard in the field. Code is available at https://github.com/Xinjie-Q/MEGA.

In this work, we aim to simultaneously enhance the effectiveness and efficiency of Mixture-of-Experts (MoE) methods. To achieve this, we propose MoE++, a general and heterogeneous MoE framework that integrates both Feed-Forward Network (FFN) and zero-computation experts. Specifically, we introduce three types of zero-computation experts: the zero expert, copy expert, and constant expert, which correspond to discard, skip, and replace operations, respectively. This design offers three key advantages: (i) **Low Computing Overhead**: Unlike the uniform mixing mechanism for all tokens within vanilla MoE, MoE++ allows each token to engage with a dynamic number of FFNs, be adjusted by constant vectors, or even skip the MoE layer entirely. (ii) **High Performance**: By enabling simple tokens to utilize fewer FFN experts, MoE++ allows more experts to focus on challenging tokens, thereby unlocking greater performance potential than vanilla MoE. (iii) **Deployment Friendly**: Given that zero-computation experts have negligible parameters, we can deploy all zero-computation experts on each GPU, eliminating the significant communication overhead and expert load imbalance associated with FFN experts distributed across different GPUs. Moreover, we leverage gating residuals, enabling each token to consider the pathway taken in the previous layer when selecting the appropriate experts. Extensive experimental results demonstrate that MoE++ achieves better performance while delivering 1.1$\sim$2.1$\times$ expert forward throughput compared to a vanilla MoE model of the same size, which lays a solid foundation for developing advanced and efficient MoE-related models.

Large language models (LLMs) like GPT-4, DeepSeek-R1, and ReasonFlux have shown significant improvements in various reasoning tasks. However, smaller LLMs still struggle with complex mathematical reasoning because they fail to effectively identify and correct reasoning errors. Recent reflection-based methods aim to address these issues by enabling self-reflection and self-correction, but they still face challenges in independently detecting errors in their reasoning steps. To overcome these limitations, we propose SuperCorrect, a novel two-stage framework that uses a large teacher model to supervise and correct both the reasoning and reflection processes of a smaller student model. In the first stage, we extract hierarchical high-level and detailed thought templates from the teacher model to guide the student model in eliciting more fine-grained reasoning thoughts. In the second stage, we introduce cross-model collaborative direct preference optimization (DPO) to enhance the self-correction abilities of the student model by following the teacher's correction traces during training. This cross-model DPO approach teaches the student model to effectively locate and resolve erroneous thoughts with error-driven insights from the teacher model, breaking the bottleneck of its thoughts and acquiring new skills and knowledge to tackle challenging problems. Extensive experiments consistently demonstrate our superiority over previous methods. Notably, our SuperCorrect-7B model significantly surpasses powerful DeepSeekMath-7B by 7.8\%/5.3\% and Qwen2.5-Math-7B by 15.1\%/6.3\% on MATH/GSM8K benchmarks, achieving new SOTA performance among all 7B models. Code is available at: https://github.com/YangLing0818/SuperCorrect-llm

We present Meissonic, which elevates non-autoregressive text-to-image Masked Image Modeling (MIM) to a level comparable with state-of-the-art diffusion models like SDXL. By incorporating a comprehensive suite of architectural innovations, advanced positional encoding strategies, and optimized sampling conditions, Meissonic substantially improves MIM's performance and efficiency. Additionally, we leverage high-quality training data, integrate micro-conditions informed by human preference scores, and employ feature compression layers to further enhance image fidelity and resolution. Our model not only matches but often exceeds the performance of existing methods in generating high-quality, high-resolution images. Extensive experiments validate Meissonic’s capabilities, demonstrating its potential as a new standard in text-to-image synthesis.

3D Gaussian splatting (3DGS), known for its groundbreaking performance and efficiency, has become a dominant 3D representation and brought progress to many 3D vision tasks. However, in this work, we reveal a significant security vulnerability that has been largely overlooked in 3DGS: the computation cost of training 3DGS could be maliciously tampered by poisoning the input data. By developing an attack named Poison-splat, we reveal a novel attack surface where the adversary can poison the input images to drastically increase the computation memory and time needed for 3DGS training, pushing the algorithm towards its worst computation complexity. In extreme cases, the attack can even consume all allocable memory, leading to a Denial-of-Service (DoS) that disrupts servers, resulting in practical damages to real-world 3DGS service vendors. Such a computation cost attack is achieved by addressing a bi-level optimization problem through three tailored strategies: attack objective approximation, proxy model rendering, and optional constrained optimization. These strategies not only ensure the effectiveness of our attack but also make it difficult to defend with simple defensive measures. We hope the revelation of this novel attack surface can spark attention to this crucial yet overlooked vulnerability of 3DGS systems. Our code is available at https://github.com/jiahaolu97/poison-splat .

Multimodal Large Language Models (MLLMs) have demonstrated exceptional capabilities in processing vision-language tasks. One of the crux of MLLMs lies in vision tokenization, which involves efficiently transforming input visual signals into feature representations that are most beneficial for LLMs. However, existing vision tokenizers, essential for semantic alignment between vision and language, remain problematic. Existing methods aggressively fragment visual input, corrupting the visual semantic integrity. To address this, this paper proposes a novel dynamic Semantic-Equivalent Vision Tokenizer (SeTok), which groups visual features into semantic units via a dynamic clustering algorithm, flexibly determining the number of tokens based on image complexity. The resulting vision tokens effectively preserve semantic integrity and capture both low-frequency and high-frequency visual features. The proposed MLLM (Setokim) equipped with SeTok significantly demonstrates superior performance across various tasks, as evidenced by our experimental results.

We introduce AGSA, an Agent-Gated Shared Autonomy framework that learns from high-level human feedback to tackle the challenges of reward-free training, safe exploration, and imperfect low-level human control. Recent human-in-the loop learning methods enable human participants to intervene a learning agent’s control and provide online demonstrations. Nonetheless, these methods rely heavily on perfect human interactions, including accurate human-monitored intervention decisions and near-optimal human demonstrations. AGSA employs a dedicated gating agent to determine when to switch control, thereby reducing the need of constant human monitoring. To obtain a precise and foreseeable gating agent, AGSA trains a long-term gating value function from human evaluative feedback on the gating agent’s intervention requests and preference feedback on pairs of human intervention trajectories. Instead of relying on potentially suboptimal human demonstrations, the learning agent is trained using control-switching signals from the gating agent. We provide theoretical insights on performance bounds that respectively describe the ability of the two agents. Experiments are conducted with both simulated and real human participants at different skill levels in challenging continuous control environments. Comparative results highlight that AGSA achieves significant improvements over previous human-in-the-loop learning methods in terms of training safety, policy performance, and user-friendliness.

Recent advances in 3D generation have been remarkable, with methods such as DreamFusion leveraging large-scale text-to-image diffusion-based models to guide 3D object generation. These methods enable the synthesis of detailed and photorealistic textured objects. However, the appearance of 3D objects produced by such text-to-3D models is often unpredictable, and it is hard for single-image-to-3D methods to deal with images lacking a clear subject, complicating the generation of appearance-controllable 3D objects from complex images. To address these challenges, we present IPDreamer, a novel method that captures intricate appearance features from complex **I**mage **P**rompts and aligns the synthesized 3D object with these extracted features, enabling high-fidelity, appearance-controllable 3D object generation. Our experiments demonstrate that IPDreamer consistently generates high-quality 3D objects that align with both the textual and complex image prompts, highlighting its promising capability in appearance-controlled, complex 3D object generation.

General virtual agents need to handle multimodal observations, master complex action spaces, and self-improve in dynamic, open-domain environments. However, existing environments are often domain-specific and require complex setups, which limits agent development and evaluation in real-world settings. As a result, current evaluations lack in-depth analyses that decompose fundamental agent capabilities. We introduce AgentStudio, a trinity of environments, tools, and benchmarks to address these issues. AgentStudio provides a lightweight, interactive environment with highly generic observation and action spaces, e.g., video observations and GUI/API actions. It integrates tools for creating online benchmark tasks, annotating GUI elements, and labeling actions in videos. Based on our environment and tools, we curate an online task suite that benchmarks both GUI interactions and function calling with efficient auto-evaluation. We also reorganize existing datasets and collect new ones using our tools to establish three datasets: GroundUI, IDMBench, and CriticBench. These datasets evaluate fundamental agent abilities, including GUI grounding, learning from videos, and success detection, pointing to the desiderata for robust, general, and open-ended virtual agents.

The Multimodal Large Language Model (MLLM) is currently experiencing rapid growth, driven by the advanced capabilities of language-based LLMs. Unlike their specialist predecessors, existing MLLMs are evolving towards a Multimodal Generalist paradigm. Initially limited to understanding multiple modalities, these models have advanced to not only comprehend but also generate across modalities. Their capabilities have expanded from coarse-grained to fine-grained multimodal understanding and from supporting singular modalities to accommodating a wide array of or even arbitrary modalities. To assess the capabilities of various MLLMs, a diverse array of benchmark test sets has been proposed. This leads to a critical question: Can we simply assume that higher performance across tasks indicates a stronger MLLM capability, bringing us closer to human-level AI? We argue that the answer is not as straightforward as it seems. In this project, we introduce an evaluation framework to delineate the capabilities and behaviors of current multimodal generalists. This framework, named General-Level, establishes 5-scale levels of MLLM performance and generality, offering a methodology to compare MLLMs and gauge the progress of existing systems towards more robust multimodal generalists and, ultimately, towards AGI (Artificial General Intelligence). Central to our framework is the use of Synergy as the evaluative criterion, categorizing capabilities based on whether MLLMs preserve synergy across comprehension and generation, as well as across multimodal interactions. To evaluate the comprehensive abilities of various generalists, we present a massive multimodal benchmark, General-Bench, which encompasses a broader spectrum of skills, modalities, formats, and capabilities, including over 700 tasks and 325,800 instances. The evaluation results that involve over 100 existing state-of-the-art MLLMs uncover the capability rankings of generalists, highlighting the challenges in reaching genuine AI. We expect this project to pave the way for future research on next-generation multimodal foundation models, providing a robust infrastructure to accelerate the realization of AGI. Project Page: https://generalist.top/, Leaderboard: https://generalist.top/leaderboard/, Benchmark: https://huggingface.co/General-Level/.

In this work, we upgrade the multi-head attention mechanism, the core of the Transformer model, to reduce computational costs while maintaining or surpassing the previous accuracy level. We show that multi-head attention can be expressed in the summation form. Drawing on the insight that not all attention heads hold equal significance, we propose Mixture-of-Head attention (MoH), a new architecture that treats attention heads as experts in the Mixture-of-Experts (MoE) mechanism. MoH has two significant advantages: First, MoH enables each token to select the appropriate attention heads, enhancing inference efficiency without compromising accuracy or increasing the number of parameters. Second, MoH replaces the standard summation in multi-head attention with a weighted summation, introducing flexibility to the attention mechanism and unlocking extra performance potential. Extensive experiments on ViT, DiT, and LLMs demonstrate that MoH outperforms multi-head attention by using only 50%$\sim$90% of the attention heads. Moreover, we demonstrate that pre-trained multi-head attention models, such as LLaMA3-8B, can be further continue-tuned into our MoH models. Notably, MoH-LLaMA3-8B achieves an average accuracy of 64.0% across 14 benchmarks, outperforming LLaMA3-8B by 2.4% by utilizing only 75% of the attention heads. We believe the proposed MoH is a promising alternative to multi-head attention and provides a strong foundation for developing advanced and efficient attention-based models.

Despite their success in specific scenarios, existing foundation agents still struggle to generalize across various virtual scenarios, mainly due to the dramatically different encapsulations of environments with manually designed observation and action spaces. To handle this issue, we propose the General Computer Control (GCC) setting to restrict foundation agents to interact with software through the most unified and standardized interface, i.e., using screenshots as input and keyboard and mouse actions as output. We introduce Cradle, a modular and flexible LMM-powered framework, as a preliminary attempt towards GCC. Enhanced by six key modules, Information Gathering, Self-Reflection, Task Inference, Skill Curation, Action Planning, and Memory, Cradle is able to understand input screenshots and output executable code for low-level keyboard and mouse control after high-level planning and information retrieval, so that Cradle can interact with any software and complete long-horizon complex tasks without relying on any built-in APIs. Experimental results show that Cradle exhibits remarkable generalizability and impressive performance across four previously unexplored commercial video games (Red Dead Redemption 2, Cities:Skylines, Stardew Valley and Dealer’s Life 2), five software applications (Chrome, Outlook, Feishu, Meitu and CapCut), and a comprehensive benchmark, OSWorld. With a unified interface to interact with any software, Cradle greatly extends the reach of foundation agents thus paving the way for generalist agents.

To learn from data collected in diverse dynamics, Imitation from Observation (IfO) methods leverage expert state trajectories based on the premise that recovering expert state distributions in other dynamics facilitates policy learning in the current one. However, Imitation Learning inherently imposes a performance upper bound of learned policies. Additionally, as the environment dynamics change, certain expert states may become inaccessible, rendering their distributions less valuable for imitation. To address this, we propose a novel framework that integrates reward maximization with IfO, employing F-distance regularized policy optimization. This framework enforces constraints on globally accessible states—those with nonzero visitation frequency across all considered dynamics—mitigating the challenge posed by inaccessible states. By instantiating F-distance in different ways, we derive two theoretical analysis and develop a practical algorithm called Accessible State Oriented Policy Regularization (ASOR). ASOR serves as a general-purpose module that can be incorporated into various RL approaches, including offline RL and off-policy RL. Extensive experiments across multiple benchmarks demonstrate ASOR’s effectiveness in enhancing state-of-the-art cross-domain policy transfer algorithms, significantly improving their performance.

This paper introduces EasyInv, an easy yet novel approach that significantly advances the field of DDIM Inversion by addressing the inherent inefficiencies and performance limitations of traditional iterative optimization methods. At the core of our EasyInv is a refined strategy for approximating inversion noise, which is pivotal for enhancing the accuracy and reliability of the inversion process. By prioritizing the initial latent state, which encapsulates rich information about the original images, EasyInv steers clear of the iterative refinement of noise items. Instead, we introduce a methodical aggregation of the latent state from the preceding time step with the current state, effectively increasing the influence of the initial latent state and mitigating the impact of noise. We illustrate that EasyInv is capable of delivering results that are either on par with or exceed those of the conventional DDIM Inversion approach, especially under conditions where the model’s precision is limited or computational resources are scarce. Concurrently, our EasyInv offers an approximate threefold enhancement regarding inference efficiency over off-the-shelf iterative optimization techniques. It can be easily combined with most existing inversion methods by only four lines of code. See code at https://github.com/potato-kitty/EasyInv.

We introduce the AR2 O Painter, an interactive intelligent system designed for real-time, highly realistic oil painting creation. This Agent can faithfully reproduce any portrait image, allowing users to visually enjoy the stroke-by-stroke painting process immersively as the artwork is completed within two minutes. It consists of two modules: the Oil Painting Stroke Sequence Planner, which performs multi-level semantic-based brushstroke sequence decomposition on portrait images, mimicking the logic of artist painting, and the Oil Painting Rendering Engine, which receives the brushstroke sequence, models the pigment via fluid dynamics, simulates its interaction with the canvas and brush, and applies a tailored PBR model with microfacet BRDF, Fresnel effects, and stroke-level geometry, enabling perceptually plausible gloss and fine-grained surface relief. To the best of our knowledge, it is the first real-time intelligent painting system to generate realistic oil paintings with high interactivity and artistic fidelity. The demo video is available at https://youtu.be/aN-W06GmnP8.

We present Pask, a proactive AI agent that provides real-time, context-aware guidance and knowledge support in audio-centric media environments. Unlike passive assistants that follow the ''you ask, I answer'' model, Pask shifts toward ''answering before asking'' by continuously monitoring live audio, anticipating user needs, and proactively offering conceptual explanations and semantic clarifications. It integrates three core components: a silent copilot for in-situ explanation, a structured knowledge base for factual grounding, and a private memory module for personalized adaptation. Pask enhances comprehension and communication in scenarios such as online learning, media consumption, and live meetings through sustained, intelligent guidance. A live demo is available at https://www.youtube.com/watch?v=ki_CKiV9Oyk.

Diffusion models have significantly improved the performance of image editing. Existing methods realize various approaches to achieve high-quality image editing, including but not limited to text control, dragging operation, and mask-and-inpainting. Among these, instruction-based editing stands out for its convenience and effectiveness in following human instructions across diverse scenarios. However, it still focuses on simple editing operations like adding, replacing, or deleting, and falls short of understanding aspects of world dynamics that convey the realistic dynamic nature in the physical world. Therefore, this work EditWorld introduces a new editing task, namely world-instructed image editing, which defines and categorizes the instructions grounded by various world scenarios. We curate a new image editing dataset with world instructions using a set of large pretrained models (e.g., GPT, Video-LLava and SDXL). To enable sufficient simulation of world dynamics for image editing, our EditWorld trains model in the curated dataset, and improves instruction-following ability with designed post-edit strategy. Extensive experiments demonstrate our method significantly outperforms existing editing methods in this new task. https://github.com/YangLing0818/EditWorld

As a form of multimedia creation, visual novel (VN) conveys engaging narratives through the integrated presentation of text, images, and music, and has shown promise across various application domains. Recent advances in generative AI have fueled interest in automating VN creation using LLMs and other foundation models. However, fully end-to-end VN creation (i.e., from user description to executable VN) remains underexplored and presents several key challenges: 1) the hallucination and limited capacity of LLMs hinder the generation of long and coherent plots; 2) current models lack effective mechanisms for ensuring cross-modal consistency between plot, visual, and audio elements. To address these issues, we propose a hierarchical end-to-end framework for automatic VN generation and assembly, which employs an outline-guided autoregressive generation mechanism that transforms high-level user prompts into coherent plots, while a vision LLM-based self-correction mechanism ensures consistency between multimedia assets and plot content. Additionally, we introduce a script validation mechanism to ensure the executable of the final VN application. Experiments demonstrate that our framework generates high-quality VN applications with coherent storylines and consistent multimedia content.

Recent advances in vision-language models (VLMs) have enabled instruction-conditioned robotic systems with improved generalization. However, most existing work focuses on reactive System 1 policies, underutilizing VLMs’ strengths in semantic reasoning and long-horizon planning. These System 2 capabilities—characterized by deliberative, goal-directed thinking—remain underexplored due to the limited temporal scale and structural complexity of current benchmarks. To address this gap, we introduce RoboCerebra, a benchmark for evaluating high-level reasoning in long-horizon robotic manipulation. RoboCerebra includes: (1) a large-scale simulation dataset with extended task horizons and diverse subtask sequences in household environments; (2) a hierarchical framework combining a high-level VLM planner with a low-level vision-language-action (VLA) controller; and (3) an evaluation protocol targeting planning, reflection, and memory through structured System 1–System 2 interaction. The dataset is constructed via a top-down pipeline, where GPT generates task instructions and decomposes them into subtask sequences. Human operators execute the subtasks in simulation, yielding high-quality trajectories with dynamic object variations. Compared to prior benchmarks, RoboCerebra features significantly longer action sequences and denser annotations. We further benchmark state-of-the-art VLMs as System 2 modules and analyze their performance across key cognitive dimensions, advancing the development of more capable and generalizable robotic planners.

Photo retouching has become integral to contemporary visual storytelling, enabling users to capture aesthetics and express creativity. While professional tools such as Adobe Lightroom offer powerful capabilities, they demand substantial expertise and manual effort. In contrast, existing AI-based solutions provide automation but often suffer from limited adjustability and poor generalization, failing to meet diverse and personalized editing needs. To bridge this gap, we introduce JarvisArt, a multi-modal large language model (MLLM)-driven agent that understands user intent, mimics the reasoning process of professional artists, and intelligently coordinates over 200 retouching tools within Lightroom. JarvisArt undergoes a two-stage training process: an initial Chain-of-Thought supervised fine-tuning to establish basic reasoning and tool-use skills, followed by Group Relative Policy Optimization for Retouching (GRPO-R) to further enhance its decision-making and tool proficiency. We also propose the Agent-to-Lightroom Protocol to facilitate seamless integration with Lightroom. To evaluate performance, we develop MMArt-Bench, a novel benchmark constructed from real-world user edits. JarvisArt demonstrates user-friendly interaction, superior generalization, and fine-grained control over both global and local adjustments, paving a new avenue for intelligent photo retouching. Notably, it outperforms GPT-4o with a 60\% improvement in average pixel-level metrics on MMArt-Bench for content fidelity, while maintaining comparable instruction-following capabilities.

This paper presents JavisGPT, the first unified multimodal large language model (MLLM) for joint audio-video (JAV) comprehension and generation. JavisGPT has a concise encoder-LLM-decoder architecture, which has a SyncFusion module for spatio-temporal audio-video fusion and synchrony-aware learnable queries to bridge a pretrained JAV-DiT generator. This design enables temporally coherent video-audio understanding and generation from multimodal instructions. We design an effective three-stage training pipeline consisting of multimodal pretraining, audio-video fine-tuning, and large-scale instruction-tuning, to progressively build multimodal comprehension and generation from existing vision-language models. For instruction tuning, we construct JavisInst-Omni, a high-quality instruction dataset with over 200K GPT-4o-curated audio-video-text dialogues that cover diverse and multi-level comprehension and generation scenarios. On JAV comprehension and generation benchmarks, our experiments show that JavisGPT outperforms existing MLLMs, particularly in complex and temporally synchronized settings.

Large language model (LLM)-powered multi-agent systems (MAS) have demonstrated cognitive and execution capabilities that far exceed those of single LLM agents, yet their capacity for self-evolution remains hampered by underdeveloped memory architectures. Upon close inspection, we are alarmed to discover that prevailing MAS memory mechanisms (1) are overly simplistic, completely disregarding the nuanced inter-agent collaboration trajectories, and (2) lack cross-trial and agent-specific customization, in stark contrast to the expressive memory developed for single agents. To bridge this gap, we introduce G-Memory, a hierarchical, agentic memory system for MAS inspired by organizational memory theory, which manages the lengthy MAS interaction via a three-tier graph hierarchy: insight, query, and interaction graphs. Upon receiving a new user query, G-Memory performs bi-directional memory traversal to retrieve both \textit{high-level, generalizable insights} that enable the system to leverage cross-trial knowledge, and \textit{fine-grained, condensed interaction trajectories} that compactly encode prior collaboration experiences. Upon task execution, the entire hierarchy evolves by assimilating new collaborative trajectories, nurturing the progressive evolution of agent teams. Extensive experiments across five benchmarks, three LLM backbones, and three popular MAS frameworks demonstrate that G-Memory improves success rates in embodied action and accuracy in knowledge QA by up to $20.89\\%$ and $10.12\\%$, respectively, without any modifications to the original frameworks.

Multimodal representation learning, with contrastive learning, plays an important role in the artificial intelligence domain. As an important subfield, video-language representation learning focuses on learning representations using global semantic interactions between pre-defined video-text pairs. However, to enhance and refine such coarse-grained global interactions, more detailed interactions are necessary for fine-grained multimodal learning. In this study, we introduce a creative approach that models video-text as game players using multivariate cooperative game theory to handle uncertainty during fine-grained semantic interactions with diverse granularity, flexible combination, and vague intensity. Specifically, we design the Hierarchical Banzhaf Interaction to simulate the finegrained correspondence between video clips and textual words from hierarchical perspectives. Furthermore, to mitigate the bias in calculations within Banzhaf Interaction, we propose reconstructing the representation through a fusion of single-modal and crossmodal components. This reconstructed representation ensures fine granularity comparable to that of the single-modal representation, while also preserving the adaptive encoding characteristics of cross-modal representation. Additionally, we extend our original structure into a flexible encoder-decoder framework, enabling the model to adapt to various downstream tasks. Extensive experiments on commonly used text-video retrieval, video-question answering, and video captioning benchmarks, with superior performance, validate the effectiveness and generalization of our method. The code is available at https://github.com/jpthu17/HBI.

Deep learning-based low-light image enhancement (LLIE) is a task of leveraging deep neural networks to enhance the image illumination while keeping the image content unchanged. From the perspective of training data, existing methods complete the LLIE task driven by one of the following three data types: paired data, unpaired data and zero-reference data. Each type of these data-driven methods has its own advantages, e.g., zero-reference data-based methods have very low requirements on training data and can meet the human needs in many scenarios. In this paper, we leverage pure Gaussian noise to complete the LLIE task, which further reduces the requirements for training data in LLIE tasks and can be used as another alternative in practical use. Specifically, we propose Noise SElf-Regression (NoiSER) without access to any task-related data, simply learns a convolutional neural network equipped with an instance-normalization layer by taking a random noise image, for each pixel, as both input and output for each training pair, and then the low-light image is fed to the trained network for predicting the normal-light image. Technically, an intuitive explanation for its effectiveness is as follows: 1) the self-regression reconstructs the contrast between adjacent pixels of the input image, 2) the instance-normalization layer may naturally remediate the overall magnitude/lighting of the input image, and 3) the assumption for each pixel enforces the output image to follow the well-known gray-world hypothesis (Buchsbaum, 1980) when the image size is big enough. Compared to current state-of-the-art LLIE methods with access to different task-related data, NoiSER is highly competitive in enhancement quality, yet with a much smaller model size, and much lower training and inference cost. In addition, the experiments also demonstrate that NoiSER has great potential in overexposure suppression and joint processing with other restoration tasks.

Offline reinforcement learning (RL) is challenged by the distributional shift problem. To tackle this issue, existing works mainly focus on designing sophisticated policy constraints between the learned policy and the behavior policy. However, these constraints are applied equally to well-performing and inferior actions through uniform sampling, which might negatively affect the learned policy. In this article, we propose offline decoupled prioritized resampling (ODPR), which designs specialized priority functions for the suboptimal policy constraint issue in offline RL and employs unique decoupled resampling for training stability. Through theoretical analysis, we show that the distinctive priority functions induce a provable improved behavior policy by modifying the distribution of the original behavior policy, and when constrained to this improved policy, a policy-constrained offline RL algorithm is likely to yield a better solution. We provide two practical implementations to balance computation and performance: one estimates priorities based on a fit value network [advantage-based ODPR (ODPR-A)] and the other utilizes trajectory returns [return-based ODPR (ODPR-R)] for quick computation. As a highly compatible plug-and-play component, ODPR is evaluated with five prevalent offline RL algorithms: behavior cloning (BC), twin delayed deep deterministic policy gradient + BC (TD3 + BC), OnestepRL, conservative Q-learning (CQL), and implicit Q-learning (IQL). Our experiments confirm that both ODPR-A and ODPR-R significantly improve performance across all baseline methods. Moreover, ODPR-A can be effective in some challenging settings, i.e., without trajectory information. Code and pretrained weights are available at https://github.com/yueyang130/ODPR.

PoolFormer is a subset of Transformer neural network with a key difference of replacing computationally demanding token mixer with pooling function. In this work, a memristor-based PoolFormer network modeling and training framework for edge-artificial intelligence (AI) applications is presented. The original PoolFormer structure is further optimized for hardware implementation on RRAM crossbar by replacing the normalization operation with scaling. In addition, the nonidealities of RRAM crossbar from device to array level as well as peripheral readout circuits are analyzed. By integrating these factors into one training framework, the overall neural network performance is evaluated holistically and the impact of nonidealities to the network performance can be effectively mitigated. Implemented in Python and PyTorch, a 16-block PoolFormer network is built with$64\times 64$four-level RRAM crossbar array model extracted from measurement results. The total number of the proposed Edge PoolFormer network parameters is 0.246 M, which is at least one order smaller than the conventional CNN implementation. This network achieved inference accuracy of 88.07% for CIFAR-10 image classification tasks with accuracy degradation of 1.5% compared to the ideal software model with FP32 precision weights.

Inspired by the long-range modeling ability of ViTs, large-kernel convolutions are widely studied and adopted recently to enlarge the receptive field and improve model performance, like the remarkable work ConvNeXt which employs 7 × 7 depthwise convolution. Although such depth-wise operator only consumes a few FLOPs, it largely harms the model efficiency on powerful computing devices due to the high memory access costs. For example, ConvNeXt-T has similar FLOPs with ResNet-50 but only achieves ~60% throughputs when trained on A100 GPUs with full precision. Although reducing the kernel size of ConvNeXt can improve speed, it results in significant performance degradation, which poses a challenging problem: How to speed up large-kernel-based CNN models while preserving their performance. To tackle this issue, inspired by Inceptions, we propose to decompose large-kernel depth-wise convolution into four parallel branches along channel dimension, i.e., small square kernel, two orthogonal band kernels, and an identity mapping. With this new Inception depthwise convolution, we build a series of networks, namely IncepitonNeXt, which not only enjoy high throughputs but also maintain competitive performance. For instance, InceptionNeXt-T achieves 1.6 × higher training throughputs than ConvNeX-T, as well as attains 0.2% top-1 accuracy improvement on ImageNet-1K. We antici-pate InceptionNeXt can serve as an economical baseline for future architecture design to reduce carbon footprint.

We tackle the common challenge of inter-concept visual confusion in compositional concept generation using text-guided diffusion models (TGDMs). It becomes even more pronounced in the generation of customized concepts, due to the scarcity of user-provided concept visual examples. By revisiting the two major stages leading to the success of TGDMs---1) contrastive image-language pre-training (CLIP) for text encoder that encodes visual semantics, and 2) training TGDM that decodes the textual embeddings into pixels---we point that existing customized generation methods only focus on fine-tuning the second stage while overlooking the first one. To this end, we propose a simple yet effective solution called CLIF: contrastive image-language fine-tuning. Specifically, given a few samples of customized concepts, we obtain non-confusing textual embeddings of a concept by fine-tuning CLIP via contrasting a concept and the over-segmented visual regions of other concepts. Experimental results demonstrate the effectiveness of CLIF in preventing the confusion of multi-customized concept generation. Project page: https://clif-official.github.io/clif.

For multimodal LLMs, the synergy of visual comprehension (textual output) and generation (visual output) presents an ongoing challenge. This is due to a conflicting objective: for comprehension, an MLLM needs to abstract the visuals; for generation, it needs to preserve the visuals as much as possible. Thus, the objective is a dilemma for visual-tokens. To resolve the conflict, we propose encoding images into morph-tokens to serve a dual purpose: for comprehension, they act as visual prompts instructing MLLM to generate texts; for generation, they take on a different, non-conflicting role as complete visual-tokens for image reconstruction, where the missing visual cues are recovered by the MLLM. Extensive experiments show that morph-tokens can achieve a new SOTA for multimodal comprehension and generation simultaneously. Our project is available at https://github.com/DCDmllm/MorphTokens.

Artificial intelligence (AI) encompasses knowledge acquisition and real-world grounding across various modalities, including language, visual, auditory, and sensory data. Multimodal large language models (MLLMs) have thus recently garnered growing interest in both academia and industry, showing an unprecedented trend to achieve human-level AI. This tutorial aims to deliver a comprehensive review of cutting-edge research in MLLMs, focusing on three key areas: MLLM architecture design, instructional learning, and multimodal reasoning of MLLMs. We will explore technical advancements, synthesize key challenges, and discuss potential avenues for future research. All the resources and materials will be made available online. https://mllm2024.github.io/ACM-MM2024

Generative AI has revolutionized multimedia, leading to groundbreaking developments in content creation, interactive experiences, and personalized media. This panel delves into the transformative potential of generative AI in academic and industrial sectors, exploring its future applications and connections to emerging techniques. Additionally, the panel will address newly identified opportunities and challenges from both technical and ethical perspectives, highlighting the importance of responsible AI development. Bringing together leading experts from universities, research institutions, and industry, this panel aims to foster discussion and debate among participants. We invite everyone to join and contribute to this critical and promising area of research in the multimedia community.

Domain generalization (DG) aims at solving distribution shift problems in various scenes. Existing approaches are based on Convolution Neural Networks (CNNs) or Vision Transformers (ViTs), which suffer from limited receptive fields or quadratic complexity issues. Mamba, as an emerging state space model (SSM), possesses superior linear complexity and global receptive fields. Despite this, it can hardly be applied to DG to address distribution shifts, due to the hidden state issues and inappropriate scan mechanisms. In this paper, we propose a novel framework for DG, named DGMamba, that excels in strong generalizability toward unseen domains and meanwhile has the advantages of global receptive fields, and efficient linear complexity. Our DGMamba compromises two core components: Hidden State Suppressing (HSS) and Semantic-aware Patch Refining (SPR). In particular, HSS is introduced to mitigate the influence of hidden states associated with domain-specific features during output prediction. SPR strives to encourage the model to concentrate more on objects rather than context, consisting of two designs: Prior-Free Scanning (PFS), and Domain Context Interchange (DCI). Concretely, PFS aims to shuffle the non-semantic patches within images, creating more flexible and effective sequences from images, and DCI is designed to regularize Mamba with the combination of mismatched non-semantic and semantic information by fusing patches among domains. Extensive experiments on four commonly used DG benchmarks demonstrate that the proposed DGMamba achieves remarkably superior results to state-of-the-art models. The code will be made publicly available at https://github.com/longshaocong/DGMamba.

Recent developments of vision large language models (LLMs) have seen remarkable progress, yet still encounter challenges towards multimodal generalists, such as coarse-grained instance-level understanding, lack of unified support for both images and videos, and insufficient coverage across various vision tasks. In this paper we present Vitron, a universal pixel-level vision LLM designed for comprehensive understanding, generating, segmenting, and editing of both static images and dynamic videos. Building on top of an LLM backbone, Vitron incorporates encoders for images, videos, and pixel-level regional visuals within its frontend modules, while employing state-of-the-art visual specialists as its backend, via which Vitron supports a spectrum of vision end tasks, spanning visual comprehension to visual generation, from low level to high level. To ensure an effective and precise message passing from LLM to backend modules for function invocation, we propose a novel hybrid method by simultaneously integrating discrete textual instructions and continuous signal embeddings. Further, we design various pixel-level spatiotemporal vision-language alignment learning for Vitron to reach the best fine-grained visual capability. Finally, a cross-task synergy module is advised to learn to maximize the task-invariant fine-grained visual features, enhancing the synergy between different visual tasks. Demonstrated over 12 visual tasks and evaluated across 22 datasets, Vitron showcases its extensive capabilities in the four main vision task clusters. Overall, this work illuminates the great potential of developing a more unified multimodal generalist.

We propose a novel method, \textbf{TwinAct}, to tackle the challenge of decoupling actions and actors in order to customize the text-guided diffusion models (TGDMs) for few-shot action image generation. TwinAct addresses the limitations of existing methods that struggle to decouple actions from other semantics (e.g., the actor's appearance) due to the lack of an effective inductive bias with few exemplar images. Our approach introduces a common action space, which is a textual embedding space focused solely on actions, enabling precise customization without actor-related details. Specifically, TwinAct involves three key steps: 1) Building common action space based on a set of representative action phrases; 2) Imitating the customized action within the action space; and 3) Generating highly adaptable customized action images in diverse contexts with action similarity loss. To comprehensively evaluate TwinAct, we construct a novel benchmark, which provides sample images with various forms of actions. Extensive experiments demonstrate TwinAct's superiority in generating accurate, context-independent customized actions while maintaining the identity consistency of different subjects, including animals, humans, and even customized actors.

Recent 3D large reconstruction models (LRMs) can generate high-quality 3D content in sub-seconds by integrating multi-view diffusion models with scalable multi-view reconstructors. Current works further leverage 3D Gaussian Splatting as 3D representation for improved visual quality and rendering efficiency. However, we observe that existing Gaussian reconstruction models often suffer from multi-view inconsistency and blurred textures. We attribute this to the compromise of multi-view information propagation in favor of adopting powerful yet computationally intensive architectures (\eg, Transformers). To address this issue, we introduce MVGamba, a general and lightweight Gaussian reconstruction model featuring a multi-view Gaussian reconstructor based on the RNN-like State Space Model (SSM). Our Gaussian reconstructor propagates causal context containing multi-view information for cross-view self-refinement while generating a long sequence of Gaussians for fine-detail modeling with linear complexity. With off-the-shelf multi-view diffusion models integrated, MVGamba unifies 3D generation tasks from a single image, sparse images, or text prompts. Extensive experiments demonstrate that MVGamba outperforms state-of-the-art baselines in all 3D content generation scenarios with approximately only $0.1\times$ of the model size. The codes are available at \url{https://github.com/SkyworkAI/MVGamba}.

Large language models (LLMs) have achieved impressive performance across various natural language benchmarks, prompting a continual need to curate more difficult datasets for larger LLMs, which is costly and time-consuming. In this paper, we propose to automate dataset updating and provide systematical analysis regarding its effectiveness in dealing with benchmark leakage issue, difficulty control, and stability. Thus, once current benchmark has been mastered or leaked, we can update it for timely and reliable evaluation. There are two updating strategies: 1) mimicking strategy to generate similar samples based on original data, preserving stylistic and contextual essence, and 2) extending strategy that further expands existing samples at varying cognitive levels by adapting Bloom’s taxonomy of educational objectives. Extensive experiments on updated MMLU and BIG-Bench demonstrate the stability of the proposed strategies and find that the mimicking strategy can effectively alleviate issues of overestimation from benchmark leakage. In cases where the efficient mimicking strategy fails, our extending strategy still shows promising results. Additionally, by controlling the difficulty, we can better discern the models’ performance and enable fine-grained analysis — neither too difficult nor too easy an exam can fairly judge students’ learning status. To the best of our knowledge, we are the first to automate updating benchmarks for reliable and timely evaluation. Our demo leaderboard can be found at https://yingjiahao14.github.io/Automating-DatasetUpdates/.

Current universal segmentation methods demonstrate strong capabilities in pixel-level image and video understanding. However, they lack reasoning abilities and cannot be controlled via text instructions. In contrast, large vision-language multimodal models exhibit powerful vision-based conversation and reasoning capabilities but lack pixel-level understanding and have difficulty accepting visual prompts for flexible user interaction. This paper proposes OMG-LLaVA, a new and elegant framework combining powerful pixel-level vision understanding with reasoning abilities. It can accept various visual and text prompts for flexible user interaction. Specifically, we use a universal segmentation method as the visual encoder, integrating image information, perception priors, and visual prompts into visual tokens provided to the LLM. The LLM is responsible for understanding the user's text instructions and providing text responses and pixel-level segmentation results based on the visual information. We propose perception prior embedding to better integrate perception priors with image features. OMG-LLaVA achieves image-level, object-level, and pixel-level reasoning and understanding in a single model, matching or surpassing the performance of specialized methods on multiple benchmarks. Rather than using LLM to connect each specialist, our work aims at end-to-end training on one encoder, one decoder, and one LLM. The code and model have been released for further research.

Training deep networks on large-scale datasets is computationally challenging. This work explores the problem of “how to accelerate adaptive gradient algorithms in a general manner", and proposes an effective Weight-decay-Integrated Nesterov acceleration (Win) to accelerate adaptive algorithms. Taking AdamW and Adam as examples, per iteration, we construct a dynamical loss that combines the vanilla training loss and a dynamic regularizer inspired by proximal point method, and respectively minimize the first- and second-order Taylor approximations of dynamical loss to update variable. This yields our Win acceleration that uses a conservative step and an aggressive step to update, and linearly combines these two updates for acceleration. Next, we extend Win into Win2 which uses multiple aggressive update steps for faster convergence. Then we apply Win and Win2 to the popular LAMB and SGD optimizers. Our transparent derivation could provide insights for other accelerated methods and their integration into adaptive algorithms. Besides, we theoretically justify the faster convergence of Win- and Win2-accelerated AdamW, Adam and LAMB to their non-accelerated counterparts. Experimental results demonstrates the faster convergence speed and superior performance of our Win- and Win2-accelerated AdamW, Adam, LAMB and SGD over their vanilla counterparts on vision classification and language modeling tasks.

While pre-training large-scale video-language models (VLMs) has shown remarkable potential for various downstream video-language tasks, existing VLMs can still suffer from certain commonly seen limitations, e.g., coarse-grained cross-modal aligning, under-modeling of temporal dynamics, detached video-language view. In this work, we target enhancing VLMs with a fine-grained structural spatio-temporal alignment learning method (namely Finsta). First of all, we represent the input texts and videos with fine-grained scene graph (SG) structures, both of which are further unified into a holistic SG (HSG) for bridging two modalities. Then, an SG-based framework is built, where the textual SG (TSG) is encoded with a graph Transformer, while the video dynamic SG (DSG) and the HSG are modeled with a novel recurrent graph Transformer for spatial and temporal feature propagation. A spatial-temporal Gaussian differential graph Transformer is further devised to strengthen the sense of the changes in objects across spatial and temporal dimensions. Next, based on the fine-grained structural features of TSG and DSG, we perform object-centered spatial alignment and predicate-centered temporal alignment respectively, enhancing the video-language grounding in both the spatiality and temporality. We design our method as a plug&play system, which can be integrated into existing well-trained VLMs for further representation augmentation, without training from scratch or relying on SG annotations in downstream applications. On 6 representative VL modeling tasks over 12 datasets in both standard and long-form video scenarios, Finsta consistently improves the existing 13 strong-performing VLMs persistently, and refreshes the current state-of-the-art end task performance significantly in both the fine-tuning and zero-shot settings.

Vision-Language Pre-Training (VLP) has demonstrated remarkable potential in aligning image and text pairs, paving the way for a wide range of cross-modal learning tasks. Nevertheless, we have observed that VLP models often fall short in terms of visual grounding and localization capabilities, which are crucial for many downstream tasks, such as visual reasoning. In response, we introduce a novel Position-guided Text Prompt (PTP) paradigm to bolster the visual grounding abilities of cross-modal models trained with VLP. In the VLP phase, PTP divides an image into N x N blocks and employs a widely-used object detector to identify objects within each block. PTP then reframes the visual grounding task as a fill-in-the-blank problem, encouraging the model to predict objects in given blocks or regress the blocks of a given object, exemplified by filling "[P]" or "[O]" in a PTP sentence such as "The block [P] has a [O]." This strategy enhances the visual grounding capabilities of VLP models, enabling them to better tackle various downstream tasks. Additionally, we integrate the seconda-order relationships between objects to further enhance the visual grounding capabilities of our proposed PTP paradigm. Incorporating PTP into several state-of-the-art VLP frameworks leads to consistently significant improvements across representative cross-modal learning model architectures and multiple benchmarks, such as zero-shot Flickr30 k Retrieval (+5.6 in average recall@1) for ViLT baseline, and COCO Captioning (+5.5 in CIDEr) for the state-of-the-art BLIP baseline. Furthermore, PTP attains comparable results with object-detector-based methods and a faster inference speed, as it discards its object detector during inference, unlike other approaches.

In deep learning, different kinds of deep networks typically need different optimizers, which have to be chosen after multiple trials, making the training process inefficient. To relieve this issue and consistently improve the model training speed across deep networks, we propose the ADAptive Nesterov momentum algorithm, Adan for short. Adan first reformulates the vanilla Nesterov acceleration to develop a new Nesterov momentum estimation (NME) method, which avoids the extra overhead of computing gradient at the extrapolation point. Then Adan adopts NME to estimate the gradient's first- and second-order moments in adaptive gradient algorithms for convergence acceleration. Besides, we prove that Adan finds an$\epsilon$-approximate first-order stationary point within$\mathcal {O}(\epsilon ^{-3.5})$stochastic gradient complexity on the non-convex stochastic problems (e.g., deep learning problems), matching the best-known lower bound. Extensive experimental results show that Adan consistently surpasses the corresponding SoTA optimizers on vision, language, and RL tasks and sets new SoTAs for many popular networks and frameworks, e.g., ResNet, ConvNext, ViT, Swin, MAE, DETR, GPT-2, Transformer-XL, and BERT. More surprisingly, Adan can use half of the training cost (epochs) of SoTA optimizers to achieve higher or comparable performance on ViT, GPT-2, MAE,etc, and also shows great tolerance to a large range of minibatch size, e.g., from 1 k to 32 k.

Existing Transformers for monocular 3D human shape and pose estimation typically have a quadratic computation and memory complexity with respect to the feature length, which hinders the exploitation of fine-grained information in high-resolution features that is beneficial for accurate reconstruction. In this work, we propose an SMPL-based Transformer framework (SMPLer) to address this issue. SMPLer incorporates two key ingredients: a decoupled attention operation and an SMPL-based target representation, which allow effective utilization of high-resolution features in the Transformer. In addition, based on these two designs, we also introduce several novel modules including a multi-scale attention and a joint-aware attention to further boost the reconstruction performance. Extensive experiments demonstrate the effectiveness of SMPLer against existing 3D human shape and pose estimation methods both quantitatively and qualitatively. Notably, the proposed algorithm achieves an MPJPE of 45.2mm on the Human3.6M dataset, improving upon the state-of-the-art approach (Lin et al., 2021) by more than 10% with fewer than one-third of the parameters.

MetaFormer, the abstracted architecture of Transformer, has been found to play a significant role in achieving competitive performance. In this paper, we further explore the capacity of MetaFormer, again, by migrating our focus away from the token mixer design: we introduce several baseline models under MetaFormer using the most basic or common mixers, and demonstrate their gratifying performance. We summarize our observations as follows: (1) MetaFormer ensures solid lower bound of performance. By merely adopting identity mapping as the token mixer, the MetaFormer model, termed IdentityFormer, achieves [Formula: see text]80% accuracy on ImageNet-1 K. (2) MetaFormer works well with arbitrary token mixers. When specifying the token mixer as even a random matrix to mix tokens, the resulting model RandFormer yields an accuracy of [Formula: see text]81%, outperforming IdentityFormer. Rest assured of MetaFormer's results when new token mixers are adopted. (3) MetaFormer effortlessly offers state-of-the-art results. With just conventional token mixers dated back five years ago, the models instantiated from MetaFormer already beat state of the art. (a) ConvFormer outperforms ConvNeXt. Taking the common depthwise separable convolutions as the token mixer, the model termed ConvFormer, which can be regarded as pure CNNs, outperforms the strong CNN model ConvNeXt. (b) CAFormer sets new record on ImageNet-1 K. By simply applying depthwise separable convolutions as token mixer in the bottom stages and vanilla self-attention in the top stages, the resulting model CAFormer sets a new record on ImageNet-1 K: it achieves an accuracy of 85.5% at 224 ×224 resolution, under normal supervised training without external data or distillation. In our expedition to probe MetaFormer, we also find that a new activation, StarReLU, reduces 71% FLOPs of activation compared with commonly-used GELU yet achieves better performance. Specifically, StarReLU is a variant of Squared ReLU dedicated to alleviating distribution shift. We expect StarReLU to find great potential in MetaFormer- like models alongside other neural networks. Code and models are available at https://github.com/sail-sg/metaformer.

AdamW modifies Adam by adding a decoupled weight decay to decay network weights per training iteration. For adaptive algorithms, this decoupled weight decay does not affect specific optimization steps, and differs from the widely used$\ell _{2}$-regularizer which changes optimization steps via changing the first- and second-order gradient moments. Despite its great practical success, for AdamW, its convergence behavior and generalization improvement over Adam and$\ell _{2}$-regularized Adam ($\ell _{2}$-Adam) remain absent yet. To solve this issue, we prove the convergence of AdamW and justify its generalization advantages over Adam and$\ell _{2}$-Adam. Specifically, AdamW provably converges but minimizes a dynamically regularized loss that combines vanilla loss and a dynamical regularization induced by decoupled weight decay, thus yielding different behaviors with Adam and$\ell _{2}$-Adam. Moreover, on both general nonconvex problems and PŁ-conditioned problems, we establish stochastic gradient complexity of AdamW to find a stationary point. Such complexity is also applicable to Adam and$\ell _{2}$-Adam, and improves their previously known complexity, especially for over-parametrized networks. Besides, we prove that AdamW enjoys smaller generalization errors than Adam and$\ell _{2}$-Adam from the Bayesian posterior aspect. This result, for the first time, explicitly reveals the benefits of decoupled weight decay in AdamW. Experimental results validate our theory.

Multi-human parsing is an image segmentation task necessitating both instance-level and fine-grained category-level information. However, prior research has typically processed these two types of information through distinct branch types and output formats, leading to inefficient and redundant frameworks. This paper introduces UniParser, which integrates instance-level and category-level representations in three key aspects: 1) we propose a unified correlation representation learning approach, allowing our network to learn instance and category features within the cosine space; 2) we unify the form of outputs of each modules as pixel-level results while supervising instance and category features using a homogeneous label accompanied by an auxiliary loss; and 3) we design a joint optimization procedure to fuse instance and category representations. By unifying instance-level and category-level output, UniParser circumvents manually designed post-processing techniques and surpasses state-of-the-art methods, achieving 49.3% AP on MHPv2.0 and 60.4% AP on CIHP. We have released our source code, pretrained models, and demos to facilitate future studies on https://github.com/cjm-sfw/Uniparser.

Single-stage models for multi-person pose estimation have garnered significant attention due to their streamlined approach in generating person position localization and body structure perception in a single pass. These two parts, however, are processed individually by existing methods, leading to suboptimal results, e.g., candidates with high confidences for person localization while poor structure estimations. To this end, we propose a simple yet effective approach, namely Structure-guided Person Localization (SPL), jointly leveraging the advantages of the two aspects to solve the multi-person pose estimation problem, with two complementary novelties. First, we propose to incorporate body structure perception to guide person position localization, consequently, we introduce the Structure-guided Center Learning (SCL) to unify the quality of the body structure perception in the displacement map with the confidence of the person existence in the center map, thus achieving more accurate keypoint position localization results even with extreme poses. Second, to facilitate the end-to-end training of SPL, we propose the efficient Agency-based Scale-adaptive Learning (ASL). Specifically, we predict an agency map of the same size as the center map, which focuses on the foreground area and can adaptively adjust the scale size for each central area with the body structure perception confidence. Comprehensive experiments on challenging benchmarks including COCO and CrowdPose clearly verify the superiority of our framework, which achieves new state-of-the-art single-stage multi-person pose estimation results. Specifically, SPL obtains 72.1 AP scores and 69.5 AP scores in COCO test-dev2017 and CrowdPose test set, respectively.

While deep generative models have empowered music generation, it remains a challenging and under-explored problem to edit an existing musical piece at fine granularity. In this paper, we propose SDMuse, a unified Stochastic Differential Music editing and generation framework, which can not only compose a whole musical piece from scratch, but also modify existing musical pieces in many ways, such as combination, continuation, inpainting, and style transferring. The proposed SDMuse follows a two-stage pipeline to achieve music generation and editing on top of a hybrid representation including pianoroll and MIDI-event. In particular, SDMuse first generates/edits pianoroll by iteratively denoising through a stochastic differential equation (SDE) based on a diffusion model generative prior, and then refines the generated pianoroll and predicts MIDI-event tokens auto-regressively. We evaluate the generated music of our method onailabs1k7pop music dataset in terms of quality and controllability on various music editing and generation tasks. Experimental results demonstrate the effectiveness of our proposed stochastic differential music editing and generation process, as well as the hybrid representations.

Studying the relationship between linear discriminant analysis (LDA) and least squares regression (LSR) is of great theoretical and practical significance. It is well-known that the two-class LDA is equivalent to an LSR problem, and directly casting multiclass LDA as an LSR problem, however, becomes more challenging. Recent study reveals that the equivalence between multiclass LDA and LSR can be established based on a special class indicator matrix, but under a mild condition which may not hold under the scenarios with low-dimensional or oversampled data. In this article, we show that the equivalence between multiclass LDA and LSR can be established based on arbitrary linearly independent class indicator vectors and without any condition. In addition, we show that LDA is also equivalent to a constrained LSR based on the data-dependent indicator vectors. It can be concluded that under exactly the same mild condition, such two regressions are both equivalent to the null space LDA method. Illuminated by the equivalence of LDA and LSR, we propose a direct LDA classifier to replace the conventional framework of LDA plus extra classifier. Extensive experiments well validate the above theoretic analysis.

The field of fashion compatibility learning has attracted great attention from both the academic and industrial communities in recent years. Many studies have been carried out for fashion compatibility prediction, collocated outfit recommendation, artificial intelligence (AI)-enabled compatible fashion design, and related topics. In particular, AI-enabled compatible fashion design can be used to synthesize compatible fashion items or outfits to improve the design experience for designers or the efficacy of recommendations for customers. However, previous generative models for collocated fashion synthesis have generally focused on the image-to-image translation between fashion items of upper and lower clothing. In this article, we propose a novel outfit generation framework, i.e., OutfitGAN, with the aim of synthesizing a set of complementary items to compose an entire outfit, given one extant fashion item and reference masks of target synthesized items. OutfitGAN includes a semantic alignment module (SAM), which is responsible for characterizing the mapping correspondence between the existing fashion items and the synthesized ones, to improve the quality of the synthesized images, and a collocation classification module (CCM), which is used to improve the compatibility of a synthesized outfit. To evaluate the performance of our proposed models, we built a large-scale dataset consisting of 20 000 fashion outfits. Extensive experimental results on this dataset show that our OutfitGAN can synthesize photo-realistic outfits and outperform the state-of-the-art methods in terms of similarity, authenticity, and compatibility measurements.

Deep learning based virtual try-on system has achieved some encouraging progress recently, but there still remain several big challenges that need to be solved, such as trying on arbitrary clothes of all types, trying on the clothes from one category to another and generating image-realistic results with few artifacts. To handle this issue, we in this article first collect a new dataset with all types of clothes, i.e., tops, bottoms, and whole clothes, each one has multiple categories with rich information of clothing characteristics such as patterns, logos, and other details. Based on this dataset, we then propose the Arbitrary Virtual Try-On Network (AVTON) that is utilized for all-type clothes, which can synthesize realistic try-on images by preserving and trading off characteristics of the target clothes and the reference person. Our approach includes three modules: (1) Limbs Prediction Module, which is utilized for predicting the human body parts by preserving the characteristics of the reference person. This is especially good for handling cross-category try-on task (e.g., long sleeves ↔ short sleeves or long pants ↔ skirts), where the exposed arms or legs with the skin colors and details can be reasonably predicted; (2) Improved Geometric Matching Module, which is designed to warp clothes according to the geometry of the target person. We improve the TPS based warping method with a compactly supported radial function (Wendland’s Ψ-function); (3) Trade-Off Fusion Module, which is to tradeoff the characteristics of the warped clothes and the reference person. This module is to make the generated try-on images look more natural and realistic based on a fine-tune symmetry of the network structure. Extensive simulations are conducted and our approach can achieve better performance compared with the state-of-the-art virtual try-on methods.

Deep reinforcement learning (RL) algorithms suffer severe performance degradation when the interaction data is scarce, which limits their real-world application. Recently, visual representation learning has been shown to be effective and promising for boosting sample efficiency in RL. These methods usually rely on contrastive learning and data augmentation to train a transition model, which is different from how the model is used in RL---performing value-based planning. Accordingly, the learned representation by these visual methods may be good for recognition but not optimal for estimating state value and solving the decision problem. To address this issue, we propose a novel method, called value-consistent representation learning (VCR), to learn representations that are directly related to decision-making. More specifically, VCR trains a model to predict the future state (also referred to as the "imagined state'') based on the current one and a sequence of actions. Instead of aligning this imagined state with a real state returned by the environment, VCR applies a Q value head on both of the states and obtains two distributions of action values. Then a distance is computed and minimized to force the imagined state to produce a similar action value prediction as that by the real state. We develop two implementations of the above idea for the discrete and continuous action spaces respectively. We conduct experiments on Atari 100k and DeepMind Control Suite benchmarks to validate their effectiveness for improving sample efficiency. It has been demonstrated that our methods achieve new state-of-the-art performance for search-free RL algorithms.

Vision-Language Pre-Training (VLP) has shown promising capabilities to align image and text pairs, facilitating a broad variety of cross-modal learning tasks. However, we observe that VLP models often lack the visual grounding/localization capability which is critical for many downstream tasks such as visual reasoning. In this work, we propose a novel Position-guided Text Prompt (PTP) paradigm to enhance the visual grounding ability of cross-modal models trained with VLP. Specifically, in the VLP phase, PTP divides the image into N x N blocks, and identifies the objects in each block through the widely used object detector in VLP. It then reformulates the visual grounding task into a fill-in-the-blank problem given a PTP by encouraging the model to predict the objects in the given blocks or regress the blocks of a given object, e.g. filling “[P]” or “[O]” in a PTP “The block [P] has a [O]”. This mechanism improves the visual grounding capability of VLP models and thus helps them better handle various downstream tasks. By introducing PTP into several state-of-the-art VLP frameworks, we observe consistently significant improvements across representative cross-modal learning model architectures and several benchmarks, e.g. zero-shot Flickr30K Retrieval (+4.8 in average recall@1) for ViLT [16] baseline, and COCO Captioning (+5.3 in CIDEr) for SOTA BLIP [19] baseline. Moreover, PTP achieves comparable results with object-detector based methods [8, 23, 45], and much faster inference speed since PTP discards its object detector for inference while the later cannot.

Few-shot image generation (FSIG) learns to generate diverse and high-fidelity images from a target domain using a few (e.g., 10) reference samples. Existing FSIG methods select, preserve and transfer prior knowledge from a source generator (pretrained on a related domain) to learn the target generator. In this work, we investigate an underexplored issue in FSIG, dubbed as incompatible knowledge transfer, which would significantly degrade the realisticness of synthetic samples. Empirical observations show that the issue stems from the least significant filters from the source generator. To this end, we propose knowledge truncation to mitigate this issue in FSIG, which is a complementary operation to knowledge preservation and is implemented by a lightweight pruning-based method. Extensive experiments show that knowledge truncation is simple and effective, consistently achieving state-of-the-art performance, including challenging setups where the source and target domains are more distant. Project Page: yunqing-me.github.io/RICK.

Recently, the topic of table pre-training has attracted considerable research interest.However, how to employ table pre-training to boost the performance of tabular prediction remains an open challenge.In this paper, we propose TAPTAP, the first attempt that leverages table pre-training to empower models for tabular prediction.After pre-training on a large corpus of real-world tabular data, TAPTAP can generate high-quality synthetic tables to support various applications on tabular data, including privacy protection, low resource regime, missing value imputation, and imbalanced classification.Extensive experiments on 12 datasets demonstrate that TAPTAP outperforms a total of 16 baselines in different scenarios.Meanwhile, it can be easily combined with various backbone models, including LightGBM, Multilayer Perceptron (MLP) and Transformer.Moreover, with the aid of table pre-training, models trained using synthetic data generated by TAPTAP can even compete with models using the original dataset on half of the experimental datasets, marking a milestone in the development of synthetic tabular data generation.The code and datasets are available at https: //github.com/ZhangTP1996/TapTap.

Despite its success in image synthesis, we observe that diffusion probabilistic models (DPMs) often lack contextual reasoning ability to learn the relations among object parts in an image, leading to a slow learning process. To solve this issue, we propose a Masked Diffusion Transformer (MDT) that introduces a mask latent modeling scheme to explicitly enhance the DPMs’ ability to contextual relation learning among object semantic parts in an image. During training, MDT operates in the latent space to mask certain tokens. Then, an asymmetric masking diffusion transformer is designed to predict masked tokens from unmasked ones while maintaining the diffusion generation process. Our MDT can reconstruct the full information of an image from its incomplete contextual input, thus enabling it to learn the associated relations among image tokens. Experimental results show that MDT achieves superior image synthesis performance, e.g., a new SOTA FID score in the ImageNet data set, and has about 3× faster learning speed than the previous SOTA DiT. The source code is released at https://github.com/sail-sg/MDT.

Existing methods of privacy-preserving action recognition (PPAR) mainly focus on frame-level (spatial) privacy removal through 2D CNNs. Unfortunately, they have two major drawbacks. First, they may compromise temporal dynamics in input videos, which are critical for accurate action recognition. Second, they are vulnerable to practical attacking scenarios where attackers probe for privacy from an entire video rather than individual frames. To address these issues, we propose a novel framework STPrivacy to perform video-level PPAR. For the first time, we introduce vision Transformers into PPAR by treating a video as a tubelet sequence, and accordingly design two complementary mechanisms, i.e., sparsification and anonymization, to remove privacy from a spatio-temporal perspective. In specific, our privacy sparsification mechanism applies adaptive token selection to abandon action-irrelevant tubelets. Then, our anonymization mechanism implicitly manipulates the remaining action-tubelets to erase privacy in the embedding space through adversarial learning. These mechanisms provide significant advantages in terms of privacy preservation for human eyes and action-privacy trade-off adjustment during deployment. We additionally contribute the first two large-scale PPAR benchmarks, VP-HMDB51 and VP-UCF101, to the community. Extensive evaluations on them, as well as two other tasks, validate the effectiveness and generalization capability of our framework.

Sliced-Wasserstein Flow (SWF) is a promising approach to nonparametric generative modeling but has not been widely adopted due to its suboptimal generative quality and lack of conditional modeling capabilities. In this work, we make two major contributions to bridging this gap. First, based on a pleasant observation that (under certain conditions) the SWF of joint distributions coincides with those of conditional distributions, we propose Conditional Sliced-Wasserstein Flow (CSWF), a simple yet effective extension of SWF that enables nonparametric conditional modeling. Second, we introduce appropriate inductive biases of images into SWF with two techniques inspired by local connectivity and multiscale representation in vision research, which greatly improve the efficiency and quality of modeling images. With all the improvements, we achieve generative performance comparable with many deep parametric generative models on both conditional and unconditional tasks in a purely nonparametric fashion, demonstrating its great potential.

It has been recognized that the data generated by the denoising diffusion probabilistic model (DDPM) improves adversarial training. After two years of rapid development in diffusion models, a question naturally arises: can better diffusion models further improve adversarial training? This paper gives an affirmative answer by employing the most recent diffusion model which has higher efficiency ($\sim 20$ sampling steps) and image quality (lower FID score) compared with DDPM. Our adversarially trained models achieve state-of-the-art performance on RobustBench using only generated data (no external datasets). Under the $\ell_\infty$-norm threat model with $\epsilon=8/255$, our models achieve $70.69\\\%$ and $42.67\\\%$ robust accuracy on CIFAR-10 and CIFAR-100, respectively, i.e. improving upon previous state-of-the-art models by $+4.58\\\%$ and $+8.03\\\%$. Under the $\ell_2$-norm threat model with $\epsilon=128/255$, our models achieve $84.86\\\%$ on CIFAR-10 ($+4.44\\\%$). These results also beat previous works that use external data. We also provide compelling results on the SVHN and TinyImageNet datasets. Our code is at https://github.com/wzekai99/DM-Improves-AT.

With the advance of language models, privacy protection is receiving more attention. Training data extraction is therefore of great importance, as it can serve as a potential tool to assess privacy leakage. However, due to the difficulty of this task, most of the existing methods are proof-of-concept and still not effective enough. In this paper, we investigate and benchmark tricks for improving training data extraction using a publicly available dataset. Because most existing extraction methods use a pipeline of generating-then-ranking, i.e., generating text candidates as potential training data and then ranking them based on specific criteria, our research focuses on the tricks for both text generation (e.g., sampling strategy) and text ranking (e.g., token-level criteria). The experimental results show that several previously overlooked tricks can be crucial to the success of training data extraction. Based on the GPT-Neo 1.3B evaluation results, our proposed tricks outperform the baseline by a large margin in most cases, providing a much stronger baseline for future research. The code is available at https://github.com/weichen-yu/LM-Extraction.

PoolFormer is a type of neural network architecture that is abstracted from Transformer where the computationally heavy token mixer module is replaced with simple pooling function. This paper presents a memristor-based PoolFormer modeling and training framework for edge-AI applications. To fit for implementation on resistive crossbar array, original PoolFormer structure is further optimized by replacing normalization operation with hardware friendly scaling operation. In addition, the non-idealities of RRAM crossbar from device to array level as well as peripheral readout circuits are also included. By incorporating these elements under a single framework for network training, their impact to the network performance can be effectively mitigated. This framework is implemented in a combination of Python and PyTorch. A 16-block PoolFormer network is designed and optimized for CIFAR-10 image classification tasks using measured$\mathbf{64}\times \mathbf{64}$RRAM crossbar array results. The total network weights are only 0.26M, which is at least one order of magnitude smaller than that of the conventional DNN implementation. When compared to the ideal model with FP64 weight bit-length, 85.86% inference accuracy is reached with only 4-level weight resolutions and less than 4% accuracy loss.

Image editing plays a vital role in computer vision field, aiming to realistically manipulate images while ensuring seamless integration. It finds numerous applications across various fields. In this work, we present EditAnything, a novel approach that empowers users with unparalleled flexibility in editing and generating image content. EditAnything introduces an array of advanced features, including cross-image dragging (e.g., try-on), region-interactive editing, controllable layout generation, and virtual character replacement. By harnessing these capabilities, users can engage in interactive and flexible editing, giving captivating outcomes that uphold the integrity of the original image. With its diverse range of tools, EditAnything caters to a wide spectrum of editing needs, pushing the boundaries of image editing and unlocking exciting new possibilities. The source code is released at https://github.com/sail-sg/EditAnything.

When hearing music, it is natural for people to dance to its rhythm. Automatic dance generation, however, is a challenging task due to the physical constraints of human motion and rhythmic alignment with target music. Conventional autoregressive methods introduce compounding errors during sampling and struggle to capture the long-term structure of dance sequences. To address these limitations, we present a novel cascaded motion diffusion model, DiffDance, designed for high-resolution, long-form dance generation. This model comprises a music-to-dance diffusion model and a sequence super-resolution diffusion model. To bridge the gap between music and motion for conditional generation, DiffDance employs a pretrained audio representation learning model to extract music embeddings and further align its embedding space to motion via contrastive loss. During training our cascaded diffusion model, we also incorporate multiple geometric losses to constrain the model outputs to be physically plausible and add a dynamic loss weight that adaptively changes over diffusion timesteps to facilitate sample diversity. Through comprehensive experiments performed on the benchmark dataset AIST++, we demonstrate that DiffDance is capable of generating realistic dance sequences that align effectively with the input music. These results are comparable to those achieved by state-of-the-art autoregressive methods.

Low-light stereo image enhancement (LLSIE) aims at improving the visual quality of stereo images captured in dark conditions. However, existing methods have shown limited success in detail recovery and illumination adjustment. This can be attributed to two main factors: 1) insufficient single-scale inter-view interaction hinders the exploitation of valuable cross-view cues; 2) lacking long-range dependency leads to the inability to deal with the spatial long-range effects caused by illumination degradation. To address these limitations, we propose a novel LLSIE model named Decoupled Cross-Scale Cross-View Interaction Network (DCI-Net). Our model introduces a key component called the Decoupled Interaction Module (DIM) designed to promote sufficient dual-view information exchange. DIM decouples the dual-view information interaction by discovering multi-scale cross-view correlations and further exploring cross-scale information flow. Furthermore, we present Spatial-channel Information Mining Block (SIMB) for intra-view feature extraction, and the benefits are twofold. One is long-range dependency capture to build spatial long-range relationship, and the other is expanded channel information refinement that enhances information flow in the channel dimension. Extensive experiments conducted on Flickr1024, KITTI 2012, KITTI 2015, and Middlebury datasets show that our method obtains better illumination adjustment and detail recovery, and achieves SOTA performance compared to other related methods.

The major challenge of offline RL is the distribution shift that appears when out-of-distribution actions are queried, which makes the policy improvement direction biased by extrapolation errors. Most existing methods address this problem by penalizing the policy or value for deviating from the behavior policy during policy improvement or evaluation. In this work, we propose a novel MISA framework to approach offline RL from the perspective of Mutual Information between States and Actions in the dataset by directly constraining the policy improvement direction. MISA constructs lower bounds of mutual information parameterized by the policy and Q-values. We show that optimizing this lower bound is equivalent to maximizing the likelihood of a one-step improved policy on the offline dataset. Hence, we constrain the policy improvement direction to lie in the data manifold. The resulting algorithm simultaneously augments the policy evaluation and improvement by adding mutual information regularizations. MISA is a general framework that unifies conservative Q-learning (CQL) and behavior regularization methods (e.g., TD3+BC) as special cases. We introduce 3 different variants of MISA, and empirically demonstrate that tighter mutual information lower bound gives better offline RL performance. In addition, our extensive experiments show MISA significantly outperforms a wide range of baselines on various tasks of the D4RL benchmark, e.g., achieving 742.9 total points on gym-locomotion tasks. Our code is attached and will be released upon publication.

Recently, diffusion models have achieved great success in generative tasks. Sampling from diffusion models is equivalent to solving the reverse diffusion stochastic differential equations (SDEs) or the corresponding probability flow ordinary differential equations (ODEs). In comparison, SDE-based solvers can generate samples of higher quality and are suited for image translation tasks like stroke-based synthesis. During inference, however, existing SDE-based solvers are severely constrained by the efficiency-effectiveness dilemma. Our investigation suggests that this is because the Gaussian assumption in the reverse transition kernel is frequently violated (even in the case of simple mixture data) given a limited number of discretization steps. To overcome this limitation, we introduce a novel class of SDE-based solvers called \emph{Gaussian Mixture Solvers (GMS)} for diffusion models. Our solver estimates the first three-order moments and optimizes the parameters of a Gaussian mixture transition kernel using generalized methods of moments in each step during sampling. Empirically, our solver outperforms numerous SDE-based solvers in terms of sample quality in image generation and stroke-based synthesis in various diffusion models, which validates the motivation and effectiveness of GMS. Our code is available at https://github.com/Guohanzhong/GMS.

In diffusion models, UNet is the most popular network backbone, since its long skip connects (LSCs) to connect distant network blocks can aggregate long-distant information and alleviate vanishing gradient. Unfortunately, UNet often suffers from unstable training in diffusion models which can be alleviated by scaling its LSC coefficients smaller. However, theoretical understandings of the instability of UNet in diffusion models and also the performance improvement of LSC scaling remain absent yet. To solve this issue, we theoretically show that the coefficients of LSCs in UNet have big effects on the stableness of the forward and backward propagation and robustness of UNet. Specifically, the hidden feature and gradient of UNet at any layer can oscillate and their oscillation ranges are actually large which explains the instability of UNet training. Moreover, UNet is also provably sensitive to perturbed input, and predicts an output distant from the desired output, yielding oscillatory loss and thus oscillatory gradient. Besides, we also observe the theoretical benefits of the LSC coefficient scaling of UNet in the stableness of hidden features and gradient and also robustness. Finally, inspired by our theory, we propose an effective coefficient scaling framework ScaleLong that scales the coefficients of LSC in UNet and better improve the training stability of UNet. Experimental results on CIFAR10, CelebA, ImageNet and COCO show that our methods are superior to stabilize training, and yield about 1.5x training acceleration on different diffusion models with UNet or UViT backbones.

Offline reinforcement learning (RL) aims to learn optimal policies from offline datasets, where the parameterization of policies is crucial but often overlooked. Recently, Diffsuion-QL significantly boosts the performance of offline RL by representing a policy with a diffusion model, whose success relies on a parametrized Markov Chain with hundreds of steps for sampling. However, Diffusion-QL suffers from two critical limitations. 1) It is computationally inefficient to forward and backward through the whole Markov chain during training. 2) It is incompatible with maximum likelihood-based RL algorithms (e.g., policy gradient methods) as the likelihood of diffusion models is intractable. Therefore, we propose efficient diffusion policy (EDP) to overcome these two challenges. EDP approximately constructs actions from corrupted ones at training to avoid running the sampling chain. We conduct extensive experiments on the D4RL benchmark. The results show that EDP can reduce the diffusion policy training time from 5 days to 5 hours on gym-locomotion tasks. Moreover, we show that EDP is compatible with various offline RL algorithms (TD3, CRR, and IQL) and achieves new state-of-the-art on D4RL by large margins over previous methods.

Recently, diffusion probabilistic models (DPMs) have achieved promising results in diverse generative tasks. A typical DPM framework includes a forward process that gradually diffuses the data distribution and a reverse process that recovers the data distribution from time-dependent data scores. In this work, we observe that the stochastic reverse process of data scores is a martingale, from which concentration bounds and the optional stopping theorem for data scores can be derived. Then, we discover a simple way for calibrating an arbitrary pretrained DPM, with which the score matching loss can be reduced and the lower bounds of model likelihood can consequently be increased. We provide general calibration guidelines under various model parametrizations. Our calibration method is performed only once and the resulting models can be used repeatedly for sampling. We conduct experiments on multiple datasets to empirically validate our proposal. Our code is available at https://github.com/thudzj/Calibrated-DPMs.

Zero-shot learning (ZSL) tackles the novel class recognition problem by transferring semantic knowledge from seen classes to unseen ones. Semantic knowledge is typically represented by attribute descriptions shared between different classes, which act as strong priors for localizing object attributes that represent discriminative region features, enabling significant and sufficient visual-semantic interaction for advancing ZSL. Existing attention-based models have struggled to learn inferior region features in a single image by solely using unidirectional attention, which ignore the transferable and discriminative attribute localization of visual features for representing the key semantic knowledge for effective knowledge transfer in ZSL. In this paper, we propose a cross attribute-guided Transformer network, termed TransZero++, to refine visual features and learn accurate attribute localization for key semantic knowledge representations in ZSL. Specifically, TransZero++ employs an attribute → visual Transformer sub-net (AVT) and a visual → attribute Transformer sub-net (VAT) to learn attribute-based visual features and visual-based attribute features, respectively. By further introducing feature-level and prediction-level semantical collaborative losses, the two attribute-guided transformers teach each other to learn semantic-augmented visual embeddings for key semantic knowledge representations via semantical collaborative learning. Finally, the semantic-augmented visual embeddings learned by AVT and VAT are fused to conduct desirable visual-semantic interaction cooperated with class semantic vectors for ZSL classification. Extensive experiments show that TransZero++ achieves the new state-of-the-art results on three golden ZSL benchmarks and on the large-scale ImageNet dataset. The project website is available at: https://shiming-chen.github.io/TransZero-pp/TransZero-pp.html.

In this paper, we present Vision Permutator, a conceptually simple and data efficient MLP-like architecture for visual recognition. By realizing the importance of the positional information carried by 2D feature representations, unlike recent MLP-like models that encode the spatial information along the flattened spatial dimensions, Vision Permutator separately encodes the feature representations along the height and width dimensions with linear projections. This allows Vision Permutator to capture long-range dependencies and meanwhile avoid the attention building process in transformers. The outputs are then aggregated in a mutually complementing manner to form expressive representations. We show that our Vision Permutators are formidable competitors to convolutional neural networks (CNNs) and vision transformers. Without the dependence on spatial convolutions or attention mechanisms, Vision Permutator achieves 81.5% top-1 accuracy on ImageNet without extra large-scale training data (e.g., ImageNet-22k) using only 25M learnable parameters, which is much better than most CNNs and vision transformers under the same model size constraint. When scaling up to 88M, it attains 83.2% top-1 accuracy, greatly improving the performance of recent state-of-the-art MLP-like networks for visual recognition. We hope this work could encourage research on rethinking the way of encoding spatial information and facilitate the development of MLP-like models. PyTorch/MindSpore/Jittor code is available at https://github.com/Andrew-Qibin/VisionPermutator.

Unsupervised domain adaption has been widely adopted in tasks with scarce annotated data. Unfortunately, mapping the target-domain distribution to the source-domain unconditionally may distort the essential structural information of the target-domain data, leading to inferior performance. To address this issue, we first propose to introduce active sample selection to assist domain adaptation regarding the semantic segmentation task. By innovatively adopting multiple anchors instead of a single centroid, both source and target domains can be better characterized as multimodal distributions, in which way more complementary and informative samples are selected from the target domain. With only a little workload to manually annotate these active samples, the distortion of the target-domain distribution can be effectively alleviated, achieving a large performance gain. In addition, a powerful semi-supervised domain adaptation strategy is proposed to alleviate the long-tail distribution problem and further improve the segmentation performance. Extensive experiments are conducted on public datasets, and the results demonstrate that the proposed approach outperforms state-of-the-art methods by large margins and achieves similar performance to the fully-supervised upperbound, i.e., 71.4% mIoU on GTA5 and 71.8% mIoU on SYNTHIA. The effectiveness of each component is also verified by thorough ablation studies.

We propose to perform video question answering (VideoQA) in a Contrastive manner via a Video Graph Transformer model (CoVGT). CoVGT's uniqueness and superiority are three-fold: 1) It proposes a dynamic graph transformer module which encodes video by explicitly capturing the visual objects, their relations and dynamics, for complex spatio-temporal reasoning. 2) It designs separate video and text transformers for contrastive learning between the video and text to perform QA, instead of multi-modal transformer for answer classification. Fine-grained video-text communication is done by additional cross-modal interaction modules. 3) It is optimized by the joint fully- and self-supervised contrastive objectives between the correct and incorrect answers, as well as the relevant and irrelevant questions respectively. With superior video encoding and QA solution, we show that CoVGT can achieve much better performances than previous arts on video reasoning tasks. Its performances even surpass those models that are pretrained with millions of external data. We further show that CoVGT can also benefit from cross-modal pretraining, yet with orders of magnitude smaller data. The results demonstrate the effectiveness and superiority of CoVGT, and additionally reveal its potential for more data-efficient pretraining.

Recently, Vision Transformers (ViTs) have been broadly explored in visual recognition. With low efficiency in encoding fine-level features, the performance of ViTs is still inferior to the state-of-the-art CNNs when trained from scratch on a midsize dataset like ImageNet. Through experimental analysis, we find it is because of two reasons: 1) the simple tokenization of input images fails to model the important local structure such as edges and lines, leading to low training sample efficiency; 2) the redundant attention backbone design of ViTs leads to limited feature richness for fixed computation budgets and limited training samples. To overcome such limitations, we present a new simple and generic architecture, termed Vision Outlooker (VOLO), which implements a novel outlook attention operation that dynamically conduct the local feature aggregation mechanism in a sliding window manner across the input image. Unlike self-attention that focuses on modeling global dependencies of local features at a coarse level, our outlook attention targets at encoding finer-level features, which is critical for recognition but ignored by self-attention. Outlook attention breaks the bottleneck of self-attention whose computation cost scales quadratically with the input spatial dimension, and thus is much more memory efficient. Compared to our Tokens-To-Token Vision Transformer (T2T-ViT), VOLO can more efficiently encode fine-level features that are essential for high-performance visual recognition. Experiments show that with only 26.6 M learnable parameters, VOLO achieves 84.2% top-1 accuracy on ImageNet-1 K without using extra training data, 2.7% better than T2T-ViT with a comparable number of parameters. When the model size is scaled up to 296 M parameters, its performance can be further improved to 87.1%, setting a new record for ImageNet-1 K classification. In addition, we also take the proposed VOLO as pretrained models and report superior performance on downstream tasks, such as semantic segmentation. Code is available at https://github.com/sail-sg/volo.

Deep long-tailed learning, one of the most challenging problems in visual recognition, aims to train well-performing deep models from a large number of images that follow a long-tailed class distribution. In the last decade, deep learning has emerged as a powerful recognition model for learning high-quality image representations and has led to remarkable breakthroughs in generic visual recognition. However, long-tailed class imbalance, a common problem in practical visual recognition tasks, often limits the practicality of deep network based recognition models in real-world applications, since they can be easily biased towards dominant classes and perform poorly on tail classes. To address this problem, a large number of studies have been conducted in recent years, making promising progress in the field of deep long-tailed learning. Considering the rapid evolution of this field, this article aims to provide a comprehensive survey on recent advances in deep long-tailed learning. To be specific, we group existing deep long-tailed learning studies into three main categories (i.e., class re-balancing, information augmentation and module improvement), and review these methods following this taxonomy in detail. Afterward, we empirically analyze several state-of-the-art methods by evaluating to what extent they address the issue of class imbalance via a newly proposed evaluation metric, i.e., relative accuracy. We conclude the survey by highlighting important applications of deep long-tailed learning and identifying several promising directions for future research.

The traditional fashion industry is heavily dependent on designers whose talent and vision have a significant impact on their innovative designs. Through taking advantage of recent advances in image-to-image translation by generative adversarial networks (GANs), marked improvement in designers’ efficiency is now possible. Considering both randomness and controllability in the design process, this article presents a novel artificial intelligence (AI)-based framework for fashion design. Under this framework, a sketch-generation module which is based on latent space is firstly introduced for designing various sketches. Secondly, a rendering-generation module is proposed to learn mapping between textures and sketches to complete the task of fashion design. In order to achieve effectiveness in synthesizing semantic-aware textures on sketches, a multi-conditional feature interaction module is developed in the rendering-generation model. Moreover, two different training schemes are introduced to optimize both the sketch-generation module and the rendering-generation module. In order to evaluate the performance of our proposed models, we built a large-scale dataset which consists of 115,584 pairs of fashion item images. Experimental results demonstrate the effectiveness of our proposed method, and indicate that our model can facilitate designers’ design process by taking full advantage of the controllability of different conditions (e.g., sketch and texture) and the randomness of latent space.

Garment sewing pattern represents the intrinsic rest shape of a garment, and is the core for many applications like fashion design, virtual try-on, and digital avatars. In this work, we explore the challenging problem of recovering garment sewing patterns from daily photos for augmenting these applications. To solve the problem, we first synthesize a versatile dataset, named SewFactory, which consists of around 1M images and ground-truth sewing patterns for model training and quantitative evaluation. SewFactory covers a wide range of human poses, body shapes, and sewing patterns, and possesses realistic appearances thanks to the proposed human texture synthesis network. Then, we propose a two-level Transformer network called Sewformer, which significantly improves the sewing pattern prediction performance. Extensive experiments demonstrate that the proposed framework is effective in recovering sewing patterns and well generalizes to casually-taken human photos. Code, dataset, and pre-trained models will be released.

Transformers have shown great potential in computer vision tasks. A common belief is their attention-based token mixer module contributes most to their competence. However, recent works show the attention-based module in transformers can be replaced by spatial MLPs and the resulted models still perform quite well. Based on this observation, we hypothesize that the general architecture of the transformers, instead of the specific token mixer module, is more essential to the model's performance. To verify this, we deliberately replace the attention module in transformers with an embarrassingly simple spatial pooling operator to conduct only basic token mixing. Surprisingly, we observe that the derived model, termed as PoolFormer, achieves competitive performance on multiple computer vision tasks. For example, on ImageNet-1K, PoolFormer achieves 82.1 % top-1 accuracy, surpassing well-tuned vision transformer/MLP-like baselines DeiT-B/ResMLP-B24 by 0.3%/1.1% accuracy with 35%/52% fewer parameters and 49%/61% fewer MACs. The effectiveness of Pool-Former verifies our hypothesis and urges us to initiate the concept of “MetaFormer”, a general architecture abstracted from transformers without specifying the token mixer. Based on the extensive experiments, we argue that MetaFormer is the key player in achieving superior results for recent transformer and MLP-like models on vision tasks. This work calls for more future research dedicated to improving MetaFormer instead of focusing on the token mixer modules. Additionally, our proposed PoolFormer could serve as a starting baseline for future MetaFormer architecture design.

Low-light image enhancement (LLIE) explores how to refine the illumination and obtain natural normal-light images. Current LLIE methods mainly focus on improving the illumination, but do not consider the color consistency by reasonably incorporating color information into the LLIE process. As a result, color difference usually exists between the enhanced image and ground-truth. To address this issue, we propose a new deep color consistent network termed DCC-Net to retain the color consistency for LLIE. A new “divide and conquer” collaborative strategy is presented, which can jointly preserve color information and enhance the illumination. Specifically, the decoupling strategy of our DCC-Net decouples each color image into two main components, i.e., gray image plus color histogram. Gray image is used to generate reasonable structures and textures, and the color histogram is beneficial for preserving the color consistency. That is, they both are utilized to complete the LLIE task collaboratively. To match the color and content features, and reduce the color consistency gap between enhanced image and ground-truth, we also design a new pyramid color embedding (PCE) module, which can better embed color information into the LLIE process. Extensive experiments on six real datasets show that the enhanced images of our DCC-Net are more natural and colorful, and perform favorably against the state-of-the-art methods.

To obtain lower inference latency and less memory footprint of deep neural networks, model quantization has been widely employed in deep model deployment, by converting the floating points to low-precision integers. However, previous methods (such as quantization aware training and post training quantization) require original data for the fine-tuning or calibration of quantized model, which makes them inapplicable to the cases that original data are not accessed due to privacy or security. This gives birth to the data-free quantization method with synthetic data generation. While current data-free quantization methods still suffer from severe performance degradation when quantizing a model into lower bit, caused by the low inter-class separability of semantic features. To this end, we propose a new and effective data-free quantization method termed ClusterQ, which utilizes the feature distribution alignment for synthetic data generation. To obtain high inter-class separability of semantic features, we cluster and align the feature distribution statistics to imitate the distribution of real data, so that the performance degradation is alleviated. Moreover, we incorporate the diversity enhancement to solve class-wise mode collapse. We also employ the exponential moving average to update the centroid of each cluster for further feature distribution improvement. Extensive experiments based on different deep models (e.g., ResNet-18 and MobileNet-V2) over the ImageNet dataset demonstrate that our proposed ClusterQ model obtains state-of-the-art performance.

The trade-off between robustness and accuracy has been widely studied in the adversarial literature. Although still controversial, the prevailing view is that this trade-off is inherent, either empirically or theoretically. Thus, we dig for the origin of this trade-off in adversarial training and find that it may stem from the improperly defined robust error, which imposes an inductive bias of local invariance — an overcorrection towards smoothness. Given this, we advocate employing local equivariance to describe the ideal behavior of a robust model, leading to a self-consistent robust error named SCORE. By definition, SCORE facilitates the reconciliation between robustness and accuracy, while still handling the worst-case uncertainty via robust optimization. By simply substituting KL divergence with variants of distance metrics, SCORE can be efficiently minimized. Empirically, our models achieve top-rank performance on RobustBench under AutoAttack. Besides, SCORE provides instructive insights for explaining the overfitting phenomenon and semantic input gradients observed on robust models.

Synchronous Rain streaks and Raindrops Removal (SR^3) is a hard and challenging task, since rain streaks and raindrops are two wildly divergent real-world phenomena with different optical properties and mathematical distributions. As such, most existing data-driven deep Singe Image Deraining (SID) methods only focus on one of them. Although there are only a few existing SR^3 methods, they still suffer from blur textures and unknown noise in reality due to weak robustness and generalization ability. In this paper, we will propose a new and universal SID model with novel modules, termed Robust Attention Deraining Network (RadNet), with strong robustness and generalization ability that are reflected in two main aspects. (1) RadNet can restore different rain degenerations, including raindrops, rain streaks, or both; (2) RadNet can adapt to different data strategies, including single-type, superimposed-type, and blended-type. The generalization ability is also demonstrated by the performance of dealing with real rain images. Specifically, we first design a lightweight and robust attention module (RAM) with a universal attention mechanism for coarse rain removal, and then present a new deep refining module (DRM) with multi-scale blocks for precise rain removal. To solve the inconsistent labels of real scenario data, we also introduce a flow & warp module (FWM) into the network, which can greatly improve the performance of real scenario data via optical flow prediction and alignment. The whole process is unified in a network to ensure sufficient robustness and strong generalization ability. We evaluated the performance of our method under a variety of data strategies, and extensive experiments demonstrated that our RadNet could outperform other state-of-the-art SID methods.

Blind image deblurring is still a challenging problem due to the inherent ill-posed properties. To improve the deblurring performance, many supervised methods have been proposed. However, obtaining labeled samples from a specific distribution (or a domain) is usually expensive, and the data-driven training-based model also cannot be generalized to the blurry images in all domains. These challenges have given birth to certain unsupervised deblurring methods. However, there is a great chromatic aberration between the latent and original images, directly degrading the performance. In this paper, we therefore propose a novel unsupervised color retention network termed CRNet to perform blind motion deblurring. In addition, new concepts of blur offset estimation and adaptive blur correction are proposed to retain the color information when deblurring. As a result, unlike the previous studies, CRNet does not learn a mapping directly from the blurry image to the restored latent image, but from the blurry image to a motion offset. An adaptive blur correction operation is then performed on the blurry image to restore the latent image, thereby retaining the color information of the original image to the greatest extent. To further effectively retain the color information and extract the blur information, we also propose a new module called pyramid global blur feature perception (PGBFP). To quantitatively prove the effectiveness of our network in color retention, we propose a novel chromatic aberration quantization metrics in line with the human perception. Extensive quantitative and visualization experiments show that CRNet can obtain the state-of-the-art performance in unsupervised deblurring tasks.

Deep image inpainting methods have improved the inpainting performance greatly due to the powerful representation ability of deep learning. However, current deep inpainting networks still tend to produce unreasonable structures and blurry textures due to the ill-posed properties of the task, i.e., image inpainting is still a challenging topic. In this paper, we therefore propose a novel deep multi-resolution mutual learning (DMRML) strategy, which can fully explore the information from various resolutions. Specifically, we design a new image inpainting network, termed multi-resolution mutual network (MRM-Net), which takes the damaged images of different resolutions as input, then excavates and exploits the correlation among different resolutions to guide the image inpainting process. Technically, we designs two new modules called multi-resolution information interaction (MRII) and adaptive content enhancement (ACE). MRII aims at discovering the correlation of multiple resolutions and exchanging information, and ACE focuses on enhancing the contents using the interacted features. Note that we also present an memory preservation mechanism (MPM) to prevent from the information loss with the increasing layers. Extensive experiments on Paris Street View, Places2 and CelebA-HQ datasets demonstrate that our proposed MRM-Net can effectively recover the textures and structures, and performs favorably against other state-of-the-art methods.

Recent studies show that transformer has strong capability of building long-range dependencies, yet is incompetent in capturing high frequencies that predominantly convey local information. To tackle this issue, we present a novel and general-purpose $\textit{Inception Transformer}$, or $\textit{iFormer}$ for short, that effectively learns comprehensive features with both high- and low-frequency information in visual data. Specifically, we design an Inception mixer to explicitly graft the advantages of convolution and max-pooling for capturing the high-frequency information to transformers. Different from recent hybrid frameworks, the Inception mixer brings greater efficiency through a channel splitting mechanism to adopt parallel convolution/max-pooling path and self-attention path as high- and low-frequency mixers, while having the flexibility to model discriminative information scattered within a wide frequency range. Considering that bottom layers play more roles in capturing high-frequency details while top layers more in modeling low-frequency global information, we further introduce a frequency ramp structure, i.e., gradually decreasing the dimensions fed to the high-frequency mixer and increasing those to the low-frequency mixer, which can effectively trade-off high- and low-frequency components across different layers. We benchmark the iFormer on a series of vision tasks, and showcase that it achieves impressive performance on image classification, COCO detection and ADE20K segmentation. For example, our iFormer-S hits the top-1 accuracy of 83.4% on ImageNet-1K, much higher than DeiT-S by 3.6%, and even slightly better than much bigger model Swin-B (83.3%) with only 1/4 parameters and 1/3 FLOPs. Code and models are released at https://github.com/sail-sg/iFormer.

There has been significant progress in developing reinforcement learning (RL) training systems. Past works such as IMPALA, Apex, Seed RL, Sample Factory, and others, aim to improve the system's overall throughput. In this paper, we aim to address a common bottleneck in the RL training system, i.e., parallel environment execution, which is often the slowest part of the whole system but receives little attention. With a curated design for paralleling RL environments, we have improved the RL environment simulation speed across different hardware setups, ranging from a laptop and a modest workstation, to a high-end machine such as NVIDIA DGX-A100. On a high-end machine, EnvPool achieves one million frames per second for the environment execution on Atari environments and three million frames per second on MuJoCo environments. When running EnvPool on a laptop, the speed is 2.8x that of the Python subprocess. Moreover, great compatibility with existing RL training libraries has been demonstrated in the open-sourced community, including CleanRL, rl_games, DeepMind Acme, etc. Finally, EnvPool allows researchers to iterate their ideas at a much faster pace and has great potential to become the de facto RL environment execution engine. Example runs show that it only takes five minutes to train agents to play Atari Pong and MuJoCo Ant on a laptop. EnvPool is open-sourced at https://github.com/sail-sg/envpool.

In this paper, we address the makeup transfer and removal tasks simultaneously, which aim to transfer the makeup from a reference image to a source image and remove the makeup from the with-makeup image respectively. Existing methods have achieved much advancement in constrained scenarios, but it is still very challenging for them to transfer makeup between images with large pose and expression differences, or handle makeup details like blush on cheeks or highlight on the nose. In addition, they are hardly able to control the degree of makeup during transferring or to transfer a specified part in the input face. These defects limit the application of previous makeup transfer methods to real-world scenarios. In this work, we propose a Pose and expression robust Spatial-aware GAN (abbreviated as PSGAN++). PSGAN++ is capable of performing both detail-preserving makeup transfer and effective makeup removal. For makeup transfer, PSGAN++ uses a Makeup Distill Network (MDNet) to extract makeup information, which is embedded into spatial-aware makeup matrices. We also devise an Attentive Makeup Morphing (AMM) module that specifies how the makeup in the source image is morphed from the reference image, and a makeup detail loss to supervise the model within the selected makeup detail area. On the other hand, for makeup removal, PSGAN++ applies an Identity Distill Network (IDNet) to embed the identity information from with-makeup images into identity matrices. Finally, the obtained makeup/identity matrices are fed to a Style Transfer Network (STNet) that is able to edit the feature maps to achieve makeup transfer or removal. To evaluate the effectiveness of our PSGAN++, we collect a Makeup Transfer In the Wild (MT-Wild) dataset that contains images with diverse poses and expressions and a Makeup Transfer High-Resolution (MT-HR) dataset that contains high-resolution images. Experiments demonstrate that PSGAN++ not only achieves state-of-the-art results with fine makeup details even in cases of large pose/expression differences but also can perform partial or degree-controllable makeup transfer. Both the code and the newly collected datasets will be released at https://github.com/wtjiang98/PSGAN.

In this paper, we target at the Fine-grAined human-Centric Tracklet Segmentation (FACTS) problem, where 12 human parts, e.g., face, pants, left-leg, are segmented. To reduce the heavy and tedious labeling efforts, FACTS requires only one labeled frame per video during training. The small size of human parts and the labeling scarcity makes FACTS very challenging. Considering adjacent frames of videos are continuous and human usually do not change clothes in a short time, we explicitly consider the pixel-level and frame-level context in the proposed Temporal Context segmentation Network (TCNet). On the one hand, optical flow is on-line calculated to propagate the pixel-level segmentation results to neighboring frames. On the other hand, frame-level classification likelihood vectors are also propagated to nearby frames. By fully exploiting the pixel-level and frame-level context, TCNet indirectly uses the large amount of unlabeled frames during training and produces smooth segmentation results during inference. Experimental results on four video datasets show the superiority of TCNet over the state-of-the-arts. The newly annotated datasets can be downloaded via http://liusi-group.com/projects/FACTS for the further studies.

Vision and language understanding techniques have achieved remarkable progress, but currently it is still difficult to well handle problems involving very fine-grained details. For example, when the robot is told to "bring me the book in the girl's left hand", most existing methods would fail if the girl holds one book respectively in her left and right hand. In this work, we introduce a new task named human-centric relation segmentation (HRS), as a fine-grained case of HOI-det. HRS aims to predict the relations between the human and surrounding entities and identify the relation-correlated human parts, which are represented as pixel-level masks. For the above exemplar case, our HRS task produces results in the form of relation triplets 〈girl [left hand], hold, book 〉 and exacts segmentation masks of the book, with which the robot can easily accomplish the grabbing task. Correspondingly, we collect a new Person In Context (PIC) dataset for this new task, which contains 17,122 high-resolution images and densely annotated entity segmentation and relations, including 141 object categories, 23 relation categories and 25 semantic human parts. We also propose a Simultaneous Matching and Segmentation (SMS) framework as a solution to the HRS task. It contains three parallel branches for entity segmentation, subject object matching and human parsing respectively. Specifically, the entity segmentation branch obtains entity masks by dynamically-generated conditional convolutions; the subject object matching branch detects the existence of any relations, links the corresponding subjects and objects by displacement estimation and classifies the interacted human parts; and the human parsing branch generates the pixelwise human part labels. Outputs of the three branches are fused to produce the final HRS results. Extensive experiments on PIC and V-COCO datasets show that the proposed SMS method outperforms baselines with the 36 FPS inference speed. Notably, SMS outperforms the best performing baseline m-KERN with only 17.6 percent time cost. The dataset and code will be released at http://picdataset.com/challenge/index/.

Despite the remarkable progress in face recognition related technologies, reliably recognizing faces across ages remains a big challenge. The appearance of a human face changes substantially over time, resulting in significant intra-class variations. As opposed to current techniques for age-invariant face recognition, which either directly extract age-invariant features for recognition, or first synthesize a face that matches target age before feature extraction, we argue that it is more desirable to perform both tasks jointly so that they can leverage each other. To this end, we propose a deep Age-Invariant Model (AIM) for face recognition in the wild with three distinct novelties. First, AIM presents a novel unified deep architecture jointly performing cross-age face synthesis and recognition in a mutual boosting way. Second, AIM achieves continuous face rejuvenation/aging with remarkable photorealistic and identity-preserving properties, avoiding the requirement of paired data and the true age of testing samples. Third, effective and novel training strategies are developed for end-to-end learning of the whole deep architecture, which generates powerful age-invariant face representations explicitly disentangled from the age variation. Moreover, we construct a new large-scale Cross-Age Face Recognition (CAFR) benchmark dataset to facilitate existing efforts and push the frontiers of age-invariant face recognition research. Extensive experiments on both our CAFR dataset and several other cross-age datasets (MORPH, CACD, and FG-NET) demonstrate the superiority of the proposed AIM model over the state-of-the-arts. Benchmarking our model on the popular unconstrained face recognition datasets YTF and IJB-C additionally verifies its promising generalization ability in recognizing faces in the wild.

Few-shot segmentation aims at learning to segment query images guided by only a few annotated images from the support set. Previous methods rely on mining the feature embedding similarity across the query and the support images to achieve successful segmentation. However, these models tend to perform badly in cases where the query instances have a large variance from the support ones. To enhance model robustness against such intra-class variance, we propose a Double Recalibration Network (DRNet) with two recalibration modules, i.e., the Self-adapted Recalibration (SR) module and the Cross-attended Recalibration (CR) module. In particular, beyond learning robust feature embedding for pixel-wise comparison between support and query as in conventional methods, the DRNet further exploits semantic-aware knowledge embedded in the query image to help segment itself, which we call ‘self-adapted recalibration’. More specifically, DRNet first employs guidance from the support set to roughly predict an incomplete but correct initial object region for the query image, and then reversely uses the feature embedding extracted from the incomplete object region to segment the query image. Also, we devise a CR module to refine the feature representation of the query image by propagating the underlying knowledge embedded in the support image’s foreground to the query. Instead of foreground global pooling, we refine the response at each pixel in the query feature map by attending to all foreground pixels in the support feature map and taking the weighted average by their similarity; meanwhile, feature maps of the query image are also added back to weighted feature maps as a residual connection. Our DRNet can effectively address the intra-class variance under the few-shot setting with such two recalibration modules, and mine more accurate target regions for query images. We conduct extensive experiments on the popular benchmarks PASCAL-$5^{i}$and COCO-$20^{i}$. The DRNet with the best configuration achieves the mIoU of$\textbf {63.6}\%$and$\textbf {64.9}\%$on PASCAL-$5^{i}$and$\textbf {44.7}\%$and$\textbf {49.6}\%$on COCO-$20^{i}$for 1-shot and 5-shot settings respectively, significantly outperforming the state-of-the-arts without any bells and whistles. Code is available at:https://github.com/fangzy97/drnet.

In this paper, we technically propose an enriched prior guided framework, called Dual-constrained Deep Semi-Supervised Coupled Factorization Network (DS2CF-Net), for discovering hierarchical coupled data representation. To extract hidden deep features, DS2CF-Net is formulated as a partial-label and geometrical structure-constrained framework. Specifically, DS2CF-Net designs a deep factorization architecture using multilayers of linear transformations, which can coupled update both the basis vectors and new representations in each layer. To enable learned deep representations and coefficients to be discriminative, we also consider enriching the supervised prior by joint deep coefficients-based label prediction and then incorporate the enriched prior information as additional label and structure constraints. The label constraint can enable the intra-class samples to have same coordinate in feature space, and the structure constraint forces the coefficients in each layer to be block-diagonal so that the enriched prior using the self-expressive label propagation are more accurate. Our network also integrates the adaptive dual-graph learning to retain the local structures of both data and feature manifolds in each layer. Extensive experiments on image datasets demonstrate the effectiveness of DS2CF-Net for representation learning and clustering.

Transformers, which are popular for language modeling, have been explored for solving vision tasks recently, e.g., the Vision Transformer (ViT) for image classification. The ViT model splits each image into a sequence of tokens with fixed length and then applies multiple Transformer layers to model their global relation for classification. However, ViT achieves inferior performance to CNNs when trained from scratch on a midsize dataset like ImageNet. We find it is because: 1) the simple tokenization of input images fails to model the important local structure such as edges and lines among neighboring pixels, leading to low training sample efficiency; 2) the redundant attention backbone design of ViT leads to limited feature richness for fixed computation budgets and limited training samples. To overcome such limitations, we propose a new Tokens-To-Token Vision Transformer (T2T-VTT), which incorporates 1) a layer-wise Tokens-to-Token (T2T) transformation to progressively structurize the image to tokens by recursively aggregating neighboring Tokens into one Token (Tokens-to-Token), such that local structure represented by surrounding tokens can be modeled and tokens length can be reduced; 2) an efficient backbone with a deep-narrow structure for vision transformer motivated by CNN architecture design after empirical study. Notably, T2T-ViT reduces the parameter count and MACs of vanilla ViT by half, while achieving more than 3.0% improvement when trained from scratch on ImageNet. It also outperforms ResNets and achieves comparable performance with MobileNets by directly training on ImageNet. For example, T2T-ViT with comparable size to ResNet50 (21.5M parameters) can achieve 83.3% top1 accuracy in image resolution 384x384 on ImageNet.1

Recently, DETR [3] pioneered the solution of vision tasks with transformers, it directly translates the image feature map into the object detection result. Though effective, translating the full feature map can be costly due to redundant computation on some area like the background. In this work, we encapsulate the idea of reducing spatial redundancy into a novel poll and pool (PnP) sampling module, with which we build an end-to-end PnP-DETR architecture that adaptively allocates its computation spatially to be more efficient. Concretely, the PnP module abstracts the image feature map into fine foreground object feature vectors and a small number of coarse background contextual feature vectors. The transformer models information interaction within the fine-coarse feature space and translates the features into the detection result. Moreover, the PnP-augmented model can instantly achieve various desired trade-offs between performance and computation with a single model by varying the sampled feature length, without requiring to train multiple models as existing methods. Thus it offers greater flexibility for deployment in diverse scenarios with varying computation constraint. We further validate the generalizability of the PnP module on panoptic segmentation and the recent transformer-based image recognition model ViT [7] and show consistent efficiency gain. We believe our method makes a step for efficient visual analysis with transformers, wherein spatial redundancy is commonly observed. Code and models will be available.

Deep subspace clustering (DSC) with the auto-encoder and self-expression layer is of great concern due to encouraging performance. However, existing methods usually adopt a “single-task” strategy based on a single dataset, without considering other related tasks or data. As such, they cannot discover other useful information to improve the clustering task. Besides, the local structure preservation of the latent codes in mapping is usually ignored. In this paper, we therefore present an effective “multi-task” strategy via the self-supervised data augmentation, and propose a new end-to-end trainable Triplet Deep Subspace Clustering Network (TDSC-net). Specifically, TDSC-net firstly generates triplet data (i.e., anchor, positive and negative data) from input data by a spectral clustering module and a self-supervised data augmentation module. This can enable it to inherit the merits of self-supervised learning and multitask learning implicitly. After that, TDSC-net builds a triplet deep autoencoder network with a self-expression layer, which takes the triplet data as input, where they share the common network layers (i.e., autoencoder and self-expression layers) over the triple tasks for complementary learning and mutual supervision. A triplet loss is also included to retain the local information of deep latent codes, which will also benefit the self-expression. Furthermore, TDSC-net separates the self-expression layer from decoding process to improve the efficiency of reconstruction. Extensive results on several public datasets demonstrate the effectiveness of our triplet-task DSC strategy.

In this work, we address the task of video background music generation. Some previous works achieve effective music generation but are unable to generate melodious music specifically for a given video, and none of them considers the video-music rhythmic consistency. To generate the background music that matches the given video, we first establish the rhythmic relationships between video and background music. In particular, we connect timing, motion speed, and motion saliency from video with beat, simu-note density, and simu-note strength from music, respectively. We then propose CMT, a Controllable Music Transformer that enables the local control of the aforementioned rhythmic features, as well as the global control of the music genre and the used instrument specified by users. Objective and subjective evaluations show that the generated background music has achieved satisfactory compatibility with the input videos, and at the same time, impressive music quality.

The fine-tuning of pre-trained language models has a great success in many NLP fields. Yet, it is strikingly vulnerable to adversarial examples, e.g., word substitution attacks using only synonyms can easily fool a BERT-based sentiment analysis model. In this paper, we demonstrate that adversarial training, the prevalent defense technique, does not directly fit a conventional fine-tuning scenario, because it suffers severely from catastrophic forgetting: failing to retain the generic and robust linguistic features that have already been captured by the pre-trained model. In this light, we propose Robust Informative Fine-Tuning (RIFT), a novel adversarial fine-tuning method from an information-theoretical perspective. In particular, RIFT encourages an objective model to retain the features learned from the pre-trained model throughout the entire fine-tuning process, whereas a conventional one only uses the pre-trained weights for initialization. Experimental results show that RIFT consistently outperforms the state-of-the-arts on two popular NLP tasks: sentiment analysis and natural language inference, under different attacks across various pre-trained language models.

We present Multi-view Pose transformer (MvP) for estimating multi-person 3D poses from multi-view images. Instead of estimating 3D joint locations from costly volumetric representation or reconstructing the per-person 3D pose from multiple detected 2D poses as in previous methods, MvP directly regresses the multi-person 3D poses in a clean and efficient way, without relying on intermediate tasks. Specifically, MvP represents skeleton joints as learnable query embeddings and let them progressively attend to and reason over the multi-view information from the input images to directly regress the actual 3D joint locations. To improve the accuracy of such a simple pipeline, MvP presents a hierarchical scheme to concisely represent query embeddings of multi-person skeleton joints and introduces an input-dependent query adaptation approach. Further, MvP designs a novel geometrically guided attention mechanism, called projective attention, to more precisely fuse the cross-view information for each joint. MvP also introduces a RayConv operation to integrate the view-dependent camera geometry into the feature representations for augmenting the projective attention. We show experimentally that our MvP model outperforms the state-of-the-art methods on several benchmarks while being much more efficient. Notably, it achieves 92.3% AP25 on the challenging Panoptic dataset, improving upon the previous best approach [35] by 9.8%. MvP is general and also extendable to recovering human mesh represented by the SMPL model, thus useful for modeling multi-person body shapes. Code and models are available at https://github.com/sail-sg/mvp.

To train networks, lookahead algorithm~\cite{zhang2019lookahead} updates its fast weights $k$ times via an inner-loop optimizer before updating its slow weights once by using the latest fast weights. Any optimizer, e.g. SGD, can serve as the inner-loop optimizer, and the derived lookahead generally enjoys remarkable test performance improvement over the vanilla optimizer. But theoretical understandings on the test performance improvement of lookahead remain absent yet. To solve this issue, we theoretically justify the advantages of lookahead in terms of the excess risk error which measures the test performance. Specifically, we prove that lookahead using SGD as its inner-loop optimizer can better balance the optimization error and generalization error to achieve smaller excess risk error than vanilla SGD on (strongly) convex problems and nonconvex problems with Polyak-{\L}ojasiewicz condition which has been observed/proved in neural networks. Moreover, we show the stagewise optimization strategy~\cite{barshan2015stage} which decays learning rate several times during training can also benefit lookahead in improving its optimization and generalization errors on strongly convex problems. Finally, we propose a stagewise locally-regularized lookahead (SLRLA) algorithm which sums up the vanilla objective and a local regularizer to minimize at each stage and provably enjoys optimization and generalization improvement over the conventional (stagewise) lookahead. Experimental results on CIFAR10/100 and ImageNet testify its advantages. Codes is available at \url{https://github.com/sail-sg/SLRLA-optimizer}.

Multi-way or tensor data analysis has attracted increasing attention recently, with many important applications in practice. This article develops a tensor low-rank representation (TLRR) method, which is the first approach that can exactly recover the clean data of intrinsic low-rank structure and accurately cluster them as well, with provable performance guarantees. In particular, for tensor data with arbitrary sparse corruptions, TLRR can exactly recover the clean data under mild conditions; meanwhile TLRR can exactly verify their true origin tensor subspaces and hence cluster them accurately. TLRR objective function can be optimized via efficient convex programing with convergence guarantees. Besides, we provide two simple yet effective dictionary construction methods, the simple TLRR (S-TLRR) and robust TLRR (R-TLRR), to handle slightly and severely corrupted data respectively. Experimental results on two computer vision data analysis tasks, image/video recovery and face clustering, clearly demonstrate the superior performance, efficiency and robustness of our developed method over state-of-the-arts including the popular LRR and SSC methods.

First-order non-convex Riemannian optimization algorithms have gained recent popularity in structured machine learning problems including principal component analysis and low-rank matrix completion. The current paper presents an efficient Riemannian Stochastic Path Integrated Differential EstimatoR (R-SPIDER) algorithm to solve the finite-sum and online Riemannian non-convex minimization problems. At the core of R-SPIDER is a recursive semi-stochastic gradient estimator that can accurately estimate Riemannian gradient under not only exponential mapping and parallel transport, but also general retraction and vector transport operations. Compared with prior Riemannian algorithms, such a recursive gradient estimation mechanism endows R-SPIDER with lower computational cost in first-order oracle complexity. Specifically, for finite-sum problems with n components, R-SPIDER is proved to converge to an ϵ-approximate stationary point within [Formula: see text] stochastic gradient evaluations, beating the best-known complexity [Formula: see text]; for online optimization, R-SPIDER is shown to converge with [Formula: see text] complexity which is, to the best of our knowledge, the first non-asymptotic result for online Riemannian optimization. For the special case of gradient dominated functions, we further develop a variant of R-SPIDER with improved linear rate of convergence. Extensive experimental results demonstrate the advantage of the proposed algorithms over the state-of-the-art Riemannian non-convex optimization methods.

Domain adaptation (DA) and transfer learning with statistical property description is very important in image analysis and data classification. This article studies the domain adaptive feature representation problem for the heterogeneous data, of which both the feature dimensions and the sample distributions across domains are so different that their features cannot be matched directly. To transfer the discriminant information efficiently from the source domain to the target domain, and then enhance the classification performance for the target data, we first introduce two projection matrices specified for different domains to transform the heterogeneous features into a shared space. We then propose a joint kernel regression model to learn the regression variable, which is called feature translator in this article. The novelty focuses on the exploration of optimal experimental design (OED) to deal with the heterogeneous and nonlinear DA by seeking the covariance structured feature translators (CSFTs). An approximate and efficient method is proposed to compute the optimal data projections. Comprehensive experiments are conducted to validate the effectiveness and efficacy of the proposed model. The results show the state-of-the-art performance of our method in heterogeneous DA.

Single Image Deraining (SID) is a relatively new and still challenging topic in emerging vision applications, and most of the recently emerged deraining methods use the supervised manner depending on the ground-truth (i.e., using paired data). However, in practice it is rather common to encounter unpaired images in real deraining task. In such cases, how to remove the rain streaks in an unsupervised way will be a challenging task due to lack of constraints between images and hence suffering from low-quality restoration results. In this paper, we therefore explore the unsupervised SID issue using unpaired data, and propose a new unsupervised framework termed DerainCycleGAN for single image rain removal and generation, which can fully utilize the constrained transfer learning ability and circulatory structures of CycleGAN. In addition, we design an unsupervised rain attentive detector (UARD) for enhancing the rain information detection by paying attention to both rainy and rain-free images. Besides, we also contribute a new synthetic way of generating the rain streak information, which is different from the previous ones. Specifically, since the generated rain streaks have diverse shapes and directions, existing derianing methods trained on the generated rainy image by this way can perform much better for processing real rainy images. Extensive experimental results on synthetic and real datasets show that our DerainCycleGAN is superior to current unsupervised and semi-supervised methods, and is also highly competitive to the fully-supervised ones.

Concept Factorization (CF) and its variants may produce inaccurate representation and clustering results due to the sensitivity to noise, hard constraint on the reconstruction error, and pre-obtained approximate similarities. To improve the representation ability, a novel unsupervised Robust Flexible Auto-weighted Local-coordinate Concept Factorization (RFA-LCF) framework is proposed for clustering high-dimensional data. Specifically, RFA-LCF integrates the robust flexible CF by clean data space recovery, robust sparse local-coordinate coding, and adaptive weighting into a unified model. RFA-LCF improves the representations by enhancing the robustness of CF to noise and errors, providing a flexible constraint on the reconstruction error and optimizing the locality jointly. For robust learning, RFA-LCF clearly learns a sparse projection to recover the underlying clean data space, and then the flexible CF is performed in the projected feature space. RFA-LCF also uses a L2,1-norm based flexible residue to encode the mismatch between the recovered data and its reconstruction, and uses the robust sparse local-coordinate coding to represent data using a few nearby basis concepts. For auto-weighting, RFA-LCF jointly preserves the manifold structures in the basis concept space and new coordinate space in an adaptive manner by minimizing the reconstruction errors on clean data, anchor points and coordinates. By updating the local-coordinate preserving data, basis concepts and new coordinates alternately, the representation abilities can be potentially improved. Extensive results on public databases show that RFA-LCF delivers the state-of-the-art clustering results compared with other related methods.

Unsupervised domain adaptation (UDA) aims at reducing the distribution discrepancy when transferring knowledge from a labeled source domain to an unlabeled target domain. Previous UDA methods assume that the source and target domains share an identical label space, which is unrealistic in practice since the label information of the target domain is agnostic. This article focuses on a more realistic UDA scenario, i.e., partial domain adaptation (PDA), where the target label space is subsumed to the source label space. In the PDA scenario, the source outliers that are absent in the target domain may be wrongly matched to the target domain (technically named negative transfer), leading to performance degradation of UDA methods. This article proposes a novel target-domain-specific classifier learning-based domain adaptation (TSCDA) method. TSCDA presents a soft-weighed maximum mean discrepancy criterion to partially align feature distributions and alleviate negative transfer. Also, it learns a target-specific classifier for the target domain with pseudolabels and multiple auxiliary classifiers to further address the classifier shift. A module named peers-assisted learning is used to minimize the prediction difference between multiple target-specific classifiers, which makes the classifiers more discriminant for the target domain. Extensive experiments conducted on three PDA benchmark data sets show that TSCDA outperforms other state-of-the-art methods with a large margin, e.g., 4% and 5.6% averagely on Office-31 and Office-Home, respectively.

Scene graph generation has received increasing attention in recent years. Enhancing the predicate representations is an important entry point to this task. There are various methods to fully investigate the context of representation enhancement. In this brief, we analyze the decisive factors that can significantly affect the relation detection results. Our analysis shows that spatial correlations between objects, focused regions of objects, and global hints related to the relations have strong influences in relation prediction and contradiction elimination. Based on our analysis, we propose a hierarchical context network (HCNet) to generate a scene graph. HCNet consists of three contexts, including interaction context, depression context, and global context, which integrates information from pair, object, and graph levels. The experiments show that our method outperforms the state-of-the-art methods on the Visual Genome (VG) data set.

Human parsing is an important task in human-centric image understanding in computer vision and multimedia systems. However, most existing works on human parsing mainly tackle the single-person scenario, which deviates from real-world applications where multiple persons are present simultaneously with interaction and occlusion. To address such a challenging multi-human parsing problem, we introduce a novel multi-human parsing model named MH-Parser, which uses a graph-based generative adversarial model to address the challenges of close-person interaction and occlusion in multi-human parsing. To validate the effectiveness of the new model, we collect a new dataset named Multi-Human Parsing (MHP), which contains multiple persons with intensive person interaction and entanglement. Experiments on the new MHP dataset and existing datasets demonstrate that the proposed method is effective in addressing the multi-human parsing problem compared with existing solutions in the literature.

Low-rank coding-based representation learning is powerful for discovering and recovering the subspace structures in data, which has obtained an impressive performance; however, it still cannot obtain deep hidden information due to the essence of single-layer structures. In this article, we investigate the deep low-rank representation of images in a progressive way by presenting a novel strategy that can extend existing single-layer latent low-rank models into multiple layers. Technically, we propose a new progressive Deep Latent Low-Rank Fusion Network (DLRF-Net) to uncover deep features and the clustering structures embedded in latent subspaces. The basic idea of DLRF-Net is to progressively refine the principal and salient features in each layer from previous layers by fusing the clustering and projective subspaces, respectively, which can potentially learn more accurate features and subspaces. To obtain deep hidden information, DLRF-Net inputs shallow features from the last layer into subsequent layers. Then, it aims at recovering the hierarchical information and deeper features by respectively congregating the subspaces in each layer of the network. As such, one can also ensure the representation learning of deeper layers to remove the noise and discover the underlying clean subspaces, which will be verified by simulations. It is noteworthy that the framework of our DLRF-Net is general and is applicable to most existing latent low-rank representation models, i.e., existing latent low-rank models can be easily extended to the multilayer scenario using DLRF-Net. Extensive results on real databases show that our framework can deliver enhanced performance over other related techniques.

Neural Architecture Search (NAS) that aims to automate the procedure of architecture design has achieved promising results in many computer vision fields. In this paper, we propose an AdversarialNAS method specially tailored for Generative Adversarial Networks (GANs) to search for a superior generative model on the task of unconditional image generation. The AdversarialNAS is the first method that can search the architectures of generator and discriminator simultaneously in a differentiable manner. During searching, the designed adversarial search algorithm does not need to comput any extra metric to evaluate the performance of the searched architecture, and the search paradigm considers the relevance between the two network architectures and improves their mutual balance. Therefore, AdversarialNAS is very efficient and only takes 1 GPU day to search for a superior generative model in the proposed large search space. Experiments demonstrate the effectiveness and superiority of our method. The discovered generative model sets a new state-of-the-art FID score of 10.87 and highly competitive Inception Score of 8.74 on CIFAR-10. Its transferability is also proven by setting new state-of-the-art FID score of 26.98 and Inception score of 9.63 on STL-10. Code is at: https://github.com/chengaopro/AdversarialNAS.

In this paper, we address the makeup transfer task, which aims to transfer the makeup from a reference image to a source image. Existing methods have achieved promising progress in constrained scenarios, but transferring between images with large pose and expression differences is still challenging. Besides, they cannot realize customizable transfer that allows a controllable shade of makeup or specifies the part to transfer, which limits their applications. To address these issues, we propose Pose and expression robust Spatial-aware GAN (PSGAN). It first utilizes Makeup Distill Network to disentangle the makeup of the reference image as two spatial-aware makeup matrices. Then, Attentive Makeup Morphing module is introduced to specify how the makeup of a pixel in the source image is morphed from the reference image. With the makeup matrices and the source image, Makeup Apply Network is used to perform makeup transfer. Our PSGAN not only achieves state-of-the-art results even when large pose and expression differences exist but also is able to perform partial and shade-controllable makeup transfer. Both the code and a newly collected dataset containing facial images with various poses and expressions will be available at https://github.com/wtjiang98/PSGAN.

For subspace recovery, most existing low-rank representation (LRR) models performs in the original space in single-layer mode. As such, the deep hierarchical information cannot be learned, which may result in inaccurate recoveries for complex real data. In this paper, we explore the deep multi-subspace recovery problem by designing a multilayer architecture for latent LRR. Technically, we propose a new Multilayer Collaborative Low-Rank Representation Network model termed DeepLRR to discover deep features and deep subspaces. In each layer (>2), DeepLRR bilinearly reconstructs the data matrix by the collaborative representation with low-rank coefficients and projection matrices in the previous layer. The bilinear low-rank reconstruction of previous layer is directly fed into the next layer as the input and low-rank dictionary for representation learning, and is further decomposed into a deep principal feature part, a deep salient feature part and a deep sparse error. As such, the coherence issue can be also resolved due to the low-rank dictionary, and the robustness against noise can also be enhanced in the feature subspace. To recover the sparse errors in layers accurately, a dynamic growing strategy is used, as the noise level will become smaller for the increase of layers. Besides, a neighborhood reconstruction error is also included to encode the locality of deep salient features by deep coefficients adaptively in each layer. Extensive results on public databases show that our DeepLRR outperforms other related models for subspace discovery and clustering.

This paper presents our solution to ACM MM challenge: Large-scale Human-centric Video Analysis in Complex Events[13]; specifically, here we focus on Track3: Crowd Pose Tracking in Complex Events. Remarkable progress has been made in multi-pose training in recent years. However, how to track the human pose in crowded and complex environments has not been well addressed. We formulate the problem as several subproblems to be solved. First, we use a multi-object tracking method to assign human ID to each bounding box generated by the detection model. After that, a pose is generated to each bounding box with ID. At last, optical flow is used to take advantage of the temporal information in the videos and generate the final pose tracking result.

Video-based human pose estimation in crowed scenes is a challenging problem due to occlusion, motion blur, scale variation and viewpoint change, etc. Prior approaches always fail to deal with this problem because of (1) lacking of usage of temporal information; (2) lacking of training data in crowded scenes. In this paper, we focus on improving human pose estimation in videos of crowded scenes from the perspectives of exploiting temporal context and collecting new data. In particular, we first follow the top-down strategy to detect persons and perform single-person pose estimation for each frame. Then, we refine the frame-based pose estimation with temporal contexts deriving from the optical-flow. Specifically, for one frame, we forward the historical poses from the previous frames and backward the future poses from the subsequent frames to current frame, leading to stable and accurate human pose estimation in videos. In addition, we mine new data of similar scenes to HIE dataset from the Internet for improving the diversity of training set. In this way, our model achieves best performance on 7 out of 13 videos and 56.33 average wAP on test dataset of HIE challenge.

Compared with the widely studied Human-Object Interaction DE-Tection (HOI-DET), no effort has been devoted to its inverse problem, i.e. to generate an HOI scene image according to the given relationship triplet , to our best knowledge. We term this new task "Human-Object Interaction Image Generation" (HOI-IG). HOI-IG is a research-worthy task with great application prospects, such as online shopping, film production and interactive entertainment. In this work, we introduce an Interact-GAN to solve this challenging task. Our method is composed of two stages: (1) manipulating the posture of a given human image conditioned on a predicate. (2) merging the transformed human image and object image to one realistic scene image while satisfying the ir expected relative position and ratio. Besides, to address the large spatial misalignment issue caused by fusing two images content with reasonable spatial layout, we propose a Relation-based Spatial Transformer Network (RSTN) to adaptively process the images conditioned on their interaction. Extensive experiments on two challenging datasets demonstrate the effectiveness and superiority of our approach. We advocate for the image generation community to draw more attention to the new Human-Object Interaction Image Generation problem. To facilitate future research, our project will be released at: http://colalab.org/projects/InteractGAN.

Customizable makeup transfer, which aims to transfer the makeup from an arbitrary reference face to a source face, is widely demanded in many applications such as short video platforms and online meeting applications. However, existing methods are neither user-friendly nor sufficiently fast. In this demo, we present the first fast makeup transfer system named as Fast Pose and expression robust Spatial-Aware GAN (FPSGAN). With a novel Attentive Makeup Morphing (AMM) module, FPSGAN is robust to face pose and expression. Moreover, it can achieve shade-controllable and partial makeup, improving the system's user-friendliness. In addition, FPSGAN is light-weighted and fast. To sum up, FPSGAN is the first fast customizable makeup transfer system to enable users to beautify themselves as they like.

Neural network architecture design is playing the most important role in recent fast development of multimedia technology. In this talk, I mainly introduce the research and development efforts in designing neural networks from two orthogonal lines: 1) how these neural network models are bio-inspired, e.g. the 1x1 convolution simulates the function of cell-body of a neuron, and 2) how these models are more hardware-friendly or motivating the next-generation of AI chips, e.g. the selective convolution is expecting new design of hardware. These two lines of efforts are collaboratively enhancing the overall efficiency of multimedia systems.

Detecting and recognizing human action in videos with crowed scenes is a challenging problem due to the complex environment and diversity events. Prior works always fail to deal with this problem in two aspects: (1) lacking utilizing information of the scenes; (2) lacking training data in the crowd and complex scenes. In this paper, we focus on improving spatio-temporal action recognition by fully-utilizing the information of scenes and collecting new data. A top-down strategy is used to overcome the limitations. Specifically, we adopt a strong human detector to detect the spatial location of each frame. We then apply action recognition models to learn the spatio-temporal information from video frames on both the HIE dataset and new data with diverse scenes from the internet, which can improve the generalization ability of our model. Besides, the scenes information is extracted by the semantic segmentation model to assistant the process. As a result, our method achieved an average 26.05 wf\_mAP (ranking 1st place in the ACM MM grand challenge 2020: Human in Events).

Pre-trained language models like BERT and its variants have recently achieved impressive performance in various natural language understanding tasks. However, BERT heavily relies on the global self-attention block and thus suffers large memory footprint and computation cost. Although all its attention heads query on the whole input sequence for generating the attention map from a global perspective, we observe some heads only need to learn local dependencies, which means existence of computation redundancy. We therefore propose a novel span-based dynamic convolution to replace these self-attention heads to directly model local dependencies. The novel convolution heads, together with the rest self-attention heads, form a new mixed attention block that is more efficient at both global and local context learning. We equip BERT with this mixed attention design and build a ConvBERT model. Experiments have shown that ConvBERT significantly outperforms BERT and its variants in various downstream tasks, with lower training cost and fewer model parameters. Remarkably, ConvBERTbase model achieves 86.4 GLUE score, 0.7 higher than ELECTRAbase, using less than 1/4 training cost. Code and pre-trained models will be released.

In this paper, we consider the Tensor Robust Principal Component Analysis (TRPCA) problem, which aims to exactly recover the low-rank and sparse components from their sum. Our model is based on the recently proposed tensor-tensor product (or t-product) [14]. Induced by the t-product, we first rigorously deduce the tensor spectral norm, tensor nuclear norm, and tensor average rank, and show that the tensor nuclear norm is the convex envelope of the tensor average rank within the unit ball of the tensor spectral norm. These definitions, their relationships and properties are consistent with matrix cases. Equipped with the new tensor nuclear norm, we then solve the TRPCA problem by solving a convex program and provide the theoretical guarantee for the exact recovery. Our TRPCA model and recovery guarantee include matrix RPCA as a special case. Numerical experiments verify our results, and the applications to image recovery and background modeling problems demonstrate the effectiveness of our method.

Rain streaks, particularly in heavy rain, not only degrade visibility but also make many computer vision algorithms fail to function properly. In this paper, we address this visibility problem by focusing on single-image rain removal, even in the presence of dense rain streaks and rain-streak accumulation, which is visually similar to mist or fog. To achieve this, we introduce a new rain model and a deep learning architecture. Our rain model incorporates a binary rain map indicating rain-streak regions, and accommodates various shapes, directions, and sizes of overlapping rain streaks, as well as rain accumulation, to model heavy rain. Based on this model, we construct a multi-task deep network, which jointly learns three targets: the binary rain-streak map, rain streak layers, and clean background, which is our ultimate output. To generate features that can be invariant to rain steaks, we introduce a contextual dilated network, which is able to exploit regional contextual information. To handle various shapes and directions of overlapping rain streaks, our strategy is to utilize a recurrent process that progressively removes rain streaks. Our binary map provides a constraint and thus additional information to train our network. Extensive evaluation on real images, particularly in heavy rain, shows the effectiveness of our model and architecture.

Many existing scene parsing methods adopt Convolutional Neural Networks with receptive fields of fixed sizes and shapes, which frequently results in inconsistent predictions of large objects and invisibility of small objects. To tackle this issue, we propose perspective-adaptive convolutions to acquire receptive fields of flexible sizes and shapes during scene parsing. Through adding a new perspective regression layer, we can dynamically infer the position-adaptive perspective coefficient vectors utilized to reshape the convolutional patches. Consequently, the receptive fields can be adjusted automatically according to the various sizes and perspective deformations of the objects in scene images. Our proposed convolutions are differentiable to learn the convolutional parameters and perspective coefficients in an end-to-end way without any extra training supervision of object sizes. Furthermore, considering that the standard convolutions lack contextual information and spatial dependencies, we propose a context adaptive bias to capture both local and global contextual information through average pooling on the local feature patches and global feature maps, followed by flexible attentive summing to the convolutional results. The attentive weights are position-adaptive and context-aware, and can be learned through adding an additional context regression layer. Experiments on Cityscapes and ADE20K datasets well demonstrate the effectiveness of the proposed methods.

We propose a joint subspace recovery and enhanced locality-based robust flexible label consistent dictionary learning method called Robust Flexible Discriminative Dictionary Learning (RFDDL). The RFDDL mainly improves the data representation and classification abilities by enhancing the robust property to sparse errors and encoding the locality, reconstruction error, and label consistency more accurately. First, for the robustness to noise and sparse errors in data and atoms, the RFDDL aims at recovering the underlying clean data and clean atom subspaces jointly, and then performs DL and encodes the locality in the recovered subspaces. Second, to enable the data sampled from a nonlinear manifold to be handled potentially and obtain the accurate reconstruction by avoiding the overfitting, the RFDDL minimizes the reconstruction error in a flexible manner. Third, to encode the label consistency accurately, the RFDDL involves a discriminative flexible sparse code error to encourage the coefficients to be soft. Fourth, to encode the locality well, the RFDDL defines the Laplacian matrix over recovered atoms, includes label information of atoms in terms of intra-class compactness and inter-class separation, and associates with group sparse codes and classifier to obtain the accurate discriminative locality-constrained coefficients and classifier. The extensive results on public databases show the effectiveness of our RFDDL.

Learning to capture dependencies between spatial positions is essential to many visual tasks, especially the dense labeling problems like scene parsing. Existing methods can effectively capture long-range dependencies with self-attention mechanism while short ones by local convolution. However, there is still much gap between long-range and short-range dependencies, which largely reduces the models' flexibility in application to diverse spatial scales and relationships in complicated natural scene images. To fill such a gap, we develop a Middle-Range (MR) branch to capture middle-range dependencies by restricting self-attention into local patches. Also, we observe that the spatial regions which have large correlations with others can be emphasized to exploit long-range dependencies more accurately, and thus propose a Reweighed Long-Range (RLR) branch. Based on the proposed MR and RLR branches, we build an Omni-Range Dependencies Network (ORDNet) which can effectively capture short-, middle- and long-range dependencies. Our ORDNet is able to extract more comprehensive context information and well adapt to complex spatial variance in scene images. Extensive experiments show that our proposed ORDNet outperforms previous state-of-the-art methods on three scene parsing benchmarks including PASCAL Context, COCO Stuff and ADE20K, demonstrating the superiority of capturing omni-range dependencies in deep models for scene parsing task.

In this paper, we investigate the robust dictionary learning (DL) to discover the hybrid salient low-rank and sparse representation in a factorized compressed space. A Joint Robust Factorization and Projective Dictionary Learning (J-RFDL) model is presented. The setting of J-RFDL aims at improving the data representations by enhancing the robustness to outliers and noise in data, encoding the reconstruction error more accurately and obtaining hybrid salient coefficients with accurate reconstruction ability. Specifically, J-RFDL performs the robust representation by DL in a factorized compressed space to eliminate the negative effects of noise and outliers on the results, which can also make the DL process efficient. To make the encoding process robust to noise in data, J-RFDL clearly uses sparse L2, 1-norm that can potentially minimize the factorization and reconstruction errors jointly by forcing rows of the reconstruction errors to be zeros. To deliver salient coefficients with good structures to reconstruct given data well, J-RFDL imposes the joint low-rank and sparse constraints on the embedded coefficients with a synthesis dictionary. Based on the hybrid salient coefficients, we also extend J-RFDL for the joint classification and propose a discriminative J-RFDL model, which can improve the discriminating abilities of learnt coefficients by minimizing the classification error jointly. Extensive experiments on public datasets demonstrate that our formulations can deliver superior performance over other state-of-the-art methods.

Constrained Concept Factorization (CCF) yields the enhanced representation ability over CF by incorporating label information as additional constraints, but it cannot classify and group unlabeled data appropriately. Minimizing the difference between the original data and its reconstruction directly can enable CCF to model a small noisy perturbation, but is not robust to gross sparse errors. Besides, CCF cannot preserve the manifold structures in new representation space explicitly, especially in an adaptive manner. In this paper, we propose a joint label prediction based Robust Semi-Supervised Adaptive Concept Factorization (RS2ACF) framework. To obtain robust representation, RS2ACF relaxes the factorization to make it simultaneously stable to small entrywise noise and robust to sparse errors. To enrich prior knowledge to enhance the discrimination, RS2ACF clearly uses class information of labeled data and more importantly propagates it to unlabeled data by jointly learning an explicit label indicator for unlabeled data. By the label indicator, RS2ACF can ensure the unlabeled data of the same predicted label to be mapped into the same class in feature space. Besides, RS2ACF incorporates the joint neighborhood reconstruction error over the new representations and predicted labels of both labeled and unlabeled data, so the manifold structures can be preserved explicitly and adaptively in the representation space and label space at the same time. Owing to the adaptive manner, the tricky process of determining the neighborhood size or kernel width can be avoided. Extensive results on public databases verify that our RS2ACF can deliver state-of-the-art data representation, compared with other related methods.

As a powerful approach for exploratory data analysis, unsupervised clustering is a fundamental task in computer vision and pattern recognition. Many clustering algorithms have been developed, but most of them perform unsatisfactorily on the data with complex structures. Recently, adversarial autoencoder (AE) (AAE) shows effectiveness on tackling such data by combining AE and adversarial training, but it cannot effectively extract classification information from the unlabeled data. In this brief, we propose dual AAE (Dual-AAE) which simultaneously maximizes the likelihood function and mutual information between observed examples and a subset of latent variables. By performing variational inference on the objective function of Dual-AAE, we derive a new reconstruction loss which can be optimized by training a pair of AEs. Moreover, to avoid mode collapse, we introduce the clustering regularization term for the category variable. Experiments on four benchmarks show that Dual-AAE achieves superior performance over state-of-the-art clustering methods. In addition, by adding a reject option, the clustering accuracy of Dual-AAE can reach that of supervised CNN algorithms. Dual-AAE can also be used for disentangling style and content of images without using supervised information.

This article presents a semisupervised multilabel fully convolutional network (FCN) for hierarchical object parsing of images. We consider each object part (e.g., eye and head) as a class label and learn to assign every image pixel to multiple coherent part labels. Different from previous methods that consider part labels as independent classes, our method explicitly models the internal relationships between object parts, e.g., that a pixel highly scored for eyes should be highly scored for heads as well. Such relationships directly reflect the structure of the semantic space and thus should be respected while learning the deep representation. We achieve this objective by introducing a multilabel softmax loss function over both labeled and unlabeled images and regularizing it with two pairwise ranking constraints. The first constraint is based on a manifold assumption that image pixels being visually and spatially close to each other should be collaboratively classified as the same part label. The other constraint is used to enforce that no pixel receives significant scores from more than one label that are semantically conflicting with each other. The proposed loss function is differentiable with respect to network parameters and hence can be optimized by standard stochastic gradient methods. We evaluate the proposed method on two public image data sets for hierarchical object parsing and compare it with the alternative parsing methods. Extensive comparisons showed that our method can achieve state-of-the-art performance while using 50% less labeled training samples than the alternatives.

The choice of which clothes to wear affects how one is perceived, as well as constitutes an expression of one's personal style. Based on the recent advances in image-to-image translation by the conditional generative adversarial network (cGAN), we propose a new framework with a multidiscriminator by incorporating different types of conditional information into the discriminator of cGAN for clothing matches. In contrast with most extant frameworks under cGAN, with one generator and one discriminator, the proposed framework investigates the potential of utilizing conditional information delivered by multidiscriminators to guide the generator. Under this framework, we propose an Attribute-GAN with two discriminators and a category-attribute GAN (CA-GAN) with three discriminators. In order to evaluate the performance of our proposed models, we built a large-scale data set that consists of 19 081 pairs of collocation clothing images with 90 manually labeled attributes. Experimental results demonstrate that with supervision of the additional attribute discriminator or category discriminator, the quality of the generated clothing images by GANs is consistently improved in comparison with the state-of-the-art methods.

In this article, we propose a deep extension of sparse subspace clustering, termed deep subspace clustering with L1-norm (DSC-L1). Regularized by the unit sphere distribution assumption for the learned deep features, DSC-L1 can infer a new data affinity matrix by simultaneously satisfying the sparsity principle of SSC and the nonlinearity given by neural networks. One of the appealing advantages brought by DSC-L1 is that when original real-world data do not meet the class-specific linear subspace distribution assumption, DSC-L1 can employ neural networks to make the assumption valid with its nonlinear transformations. Moreover, we prove that our neural network could sufficiently approximate the minimizer under mild conditions. To the best of our knowledge, this could be one of the first deep-learning-based subspace clustering methods. Extensive experiments are conducted on four real-world data sets to show that the proposed method is significantly superior to 17 existing methods for subspace clustering on handcrafted features and raw data.

In this article, we propose a structured robust adaptive dictionary pair learning (RA-DPL) framework for the discriminative sparse representation (SR) learning. To achieve powerful representation ability of the available samples, the setting of RA-DPL seamlessly integrates the robust projective DPL, locality-adaptive SRs, and discriminative coding coefficients learning into a unified learning framework. Specifically, RA-DPL improves existing projective DPL in four perspectives. First, it applies a sparse l2,1-norm-based metric to encode the reconstruction error to deliver the robust projective dictionary pairs, and the l2,1-norm has the potential to minimize the error. Second, it imposes the robust l2,1-norm clearly on the analysis dictionary to ensure the sparse property of the coding coefficients rather than using the costly l0/l1-norm. As such, the robustness of the data representation and the efficiency of the learning process are jointly considered to guarantee the efficacy of our RA-DPL. Third, RA-DPL conceives a structured reconstruction weight learning paradigm to preserve the local structures of the coding coefficients within each class clearly in an adaptive manner, which encourages to produce the locality preserving representations. Fourth, it also considers improving the discriminating ability of coding coefficients and dictionary by incorporating a discriminating function, which can ensure high intraclass compactness and interclass separation in the code space. Extensive experiments show that our RA-DPL can obtain superior performance over other state of the arts.

Despite the remarkable progress in face recognition related technologies, reliably recognizing faces across ages still remains a big challenge. The appearance of a human face changes substantially over time, resulting in significant intraclass variations. As opposed to current techniques for ageinvariant face recognition, which either directly extract ageinvariant features for recognition, or first synthesize a face that matches target age before feature extraction, we argue that it is more desirable to perform both tasks jointly so that they can leverage each other. To this end, we propose a deep Age-Invariant Model (AIM) for face recognition in the wild with three distinct novelties. First, AIM presents a novel unified deep architecture jointly performing cross-age face synthesis and recognition in a mutual boosting way. Second, AIM achieves continuous face rejuvenation/aging with remarkable photorealistic and identity-preserving properties, avoiding the requirement of paired data and the true age of testing samples. Third, we develop effective and novel training strategies for end-to-end learning the whole deep architecture, which generates powerful age-invariant face representations explicitly disentangled from the age variation. Extensive experiments on several cross-age datasets (MORPH, CACD and FG-NET) demonstrate the superiority of the proposed AIM model over the state-of-the-arts. Benchmarking our model on one of the most popular unconstrained face recognition datasets IJB-C additionally verifies the promising generalizability of AIM in recognizing faces in the wild.

Globally modeling and reasoning over relations between regions can be beneficial for many computer vision tasks on both images and videos. Convolutional Neural Networks (CNNs) excel at modeling local relations by convolution operations, but they are typically inefficient at capturing global relations between distant regions and require stacking multiple convolution layers. In this work, we propose a new approach for reasoning globally in which a set of features are globally aggregated over the coordinate space and then projected to an interaction space where relational reasoning can be efficiently computed. After reasoning, relation-aware features are distributed back to the original coordinate space for down-stream tasks. We further present a highly efficient instantiation of the proposed approach and introduce the Global Reasoning unit (GloRe unit) that implements the coordinate-interaction space mapping by weighted global pooling and weighted broadcasting, and the relation reasoning via graph convolution on a small graph in interaction space. The proposed GloRe unit is lightweight, end-to-end trainable and can be easily plugged into existing CNNs for a wide range of tasks. Extensive experiments show our GloRe unit can consistently boost the performance of state-of-the-art backbone architectures, including ResNet, ResNeXt, SE-Net and DPN, for both 2D and 3D CNNs, on image classification, semantic segmentation and video action recognition task.

We investigate the high-dimensional data clustering problem by proposing a novel and unsupervised representation learning model called Robust Flexible Auto-weighted Local-coordinate Concept Factorization (RFA-LCF). RFA-LCF integrates the robust flexible CF, robust sparse local-coordinate coding and the adaptive reconstruction weighting learning into a unified model. The adaptive weighting is driven by including thejoint manifold preserving constraints on the recovered clean data, basis concepts and new representation. Specifically, our RFA-LCF uses a L2,1-norm based flexible residue to encode the mismatch between clean data and its reconstruction, and also applies the robust adaptive sparse local-coordinate coding to represent the data using a few nearby basis concepts, which can make the factorization more accurate and robust to noise. The robust flexible factorization is also performed in the recovered clean data space for enhancing representations. RFA-LCF also considers preserving the local manifold structures of clean data space, basis concept space and the new coordinate space jointly in an adaptive manner way. Extensive comparisons show that RFA-LCF can deliver enhanced clustering results.

In natural images, information is conveyed at different frequencies where higher frequencies are usually encoded with fine details and lower frequencies are usually encoded with global structures. Similarly, the output feature maps of a convolution layer can also be seen as a mixture of information at different frequencies. In this work, we propose to factorize the mixed feature maps by their frequencies, and design a novel Octave Convolution (OctConv) operation to store and process feature maps that vary spatially “slower” at a lower spatial resolution reducing both memory and computation cost. Unlike existing multi-scale methods, OctConv is formulated as a single, generic, plug-and-play convolutional unit that can be used as a direct replacement of (vanilla) convolutions without any adjustments in the network architecture. It is also orthogonal and complementary to methods that suggest better topologies or reduce channel-wise redundancy like group or depth-wise convolutions. We experimentally show that by simply replacing convolutions with OctConv, we can consistently boost accuracy for both image and video recognition tasks, while reducing memory and computational cost. An OctConv-equipped ResNet-152 can achieve 82.9% top-1 classification accuracy on ImageNet with merely 22.2 GFLOPs.

Multi-person pose estimation is a challenging problem. Existing methods are mostly two-stage based-one stage for proposal generation and the other for allocating poses to corresponding persons. However, such two-stage methods generally suffer low efficiency. In this work, we present the first single-stage model, Single-stage multi-person Pose Machine (SPM), to simplify the pipeline and lift the efficiency for multi-person pose estimation. To achieve this, we propose a novel Structured Pose Representation (SPR) that unifies person instance and body joint position representations. Based on SPR, we develop the SPM model that can directly predict structuredposesfor multiple persons in a single stage, and thus offer a more compact pipeline and attractive efficiency advantage over two-stage methods. In particular, SPR introduces the root joints to indicate different person instances and human body joint positions are encoded into their displacements w.r.t. the roots. To better predict long-range displacements for some joints, SPR is further extended to hierarchical representations. Based on SPR, SPM can efficiently perform multi-person poses estimation by simultaneously predicting root joints (location of instances) and body joint displacements via CNNs. Moreover, to demonstrate the generality of SPM, we also apply it to multi-person 3D pose estimation. Comprehensive experiments on benchmarks MPII, extended PASCAL-PersonPart, MSCOCO and CMU Panoptic clearly demonstrate the state-of-the-art efficiency of SPM for multi-person 2D/3D pose estimation, together with outstanding accuracy.

Comparing to image inpainting, image outpainting receives less attention due to two challenges in it. The first challenge is how to keep the spatial and content consistency between generated images and original input. The second challenge is how to maintain high quality in generated results, especially for multi-step generations in which generated regions are spatially far away from the initial input. To solve the two problems, we devise some innovative modules, named Skip Horizontal Connection and Recurrent Content Transfer, and integrate them into our designed encoder-decoder structure. By this design, our network can generate highly realistic outpainting prediction effectively and efficiently. Other than that, our method can generate new images with very long sizes while keeping the same style and semantic content as the given input. To test the effectiveness of the proposed architecture, we collect a new scenery dataset with diverse, complicated natural scenes. The experimental results on this dataset have demonstrated the efficacy of our proposed network.

In this paper, we address the challenging unconstrained set-based face recognition problem where each subject face is instantiated by a set of media (images and videos) instead of a single image. Naively aggregating information from all the media within a set would suffer from the large intra-set variance caused by heterogeneous factors (e.g., varying media modalities, poses and illumination) and fail to learn discriminative face representations. A novel Multi-Prototype Network (MP- Net) model is thus proposed to learn multiple prototype face representations adaptively from the media sets. Each learned prototype is representative for the subject face under certain condition in terms of pose, illumination and media modality. Instead of handcrafting the set partition for prototype learn- ing, MPNet introduces a Dense SubGraph (DSG) learning sub-net that implicitly untangles inconsistent media and learns a number of representative prototypes. Qualitative and quantitative experiments clearly demonstrate the superiority of the proposed model over state-of-the-arts.

Image retrieval in complicated scene is a challenging task that requires the comprehensive understanding of an image. In this paper, we propose a scene graph based image retrieval framework that combines the scene graph generation with image retrieval and fine tuning the searching results via a dialogue mechanism. Specifically, we proposed an image retrieval oriented scene graph generation model that takes an image and a text describing the image as inputs. The additional text input is used to control the generated scene graph. It provides information for a newly introduced attributes head to better predict the attributes and helps constructing an adjacency matrix at the same time. Graph Convolutional Network is further used to gather information among nodes for precise relation estimation. Moreover, modification on the scene graph can be done by changing the text. Our proposed approach achieves the state-of-the-art performances in both scene graph based image retrieval and scene graph generation in the Visual Genome dataset.

We address the problem of meta-learning which learns a prior over hypothesis from a sample of meta-training tasks for fast adaptation on meta-testing tasks. A particularly simple yet successful paradigm for this research is model-agnostic meta-learning (MAML). Implementation and analysis of MAML, however, can be tricky; first-order approximation is usually adopted to avoid directly computing Hessian matrix but as a result the convergence and generalization guarantees remain largely mysterious for MAML. To remedy this deficiency, in this paper we propose a minibatch proximal update based meta-learning approach for learning to efficient hypothesis transfer. The principle is to learn a prior hypothesis shared across tasks such that the minibatch risk minimization biased regularized by this prior can quickly converge to the optimal hypothesis in each training task. The prior hypothesis training model can be efficiently optimized via SGD with provable convergence guarantees for both convex and non-convex problems. Moreover, we theoretically justify the benefit of the learnt prior hypothesis for fast adaptation to new few-shot learning tasks via minibatch proximal update. Experimental results on several few-shot regression and classification tasks demonstrate the advantages of our method over state-of-the-arts.

Multi-task learning is popular in machine learning and computer vision. In multitask learning, properly modeling task relations is important for boosting the performance of jointly learned tasks. Task covariance modeling has been successfully used to model the relations of tasks but is limited to homogeneous multi-task learning. In this paper, we propose a feature based task relation modeling approach, suitable for both homogeneous and heterogeneous multi-task learning. First, we propose a new metric to quantify the relations between tasks. Based on the quantitative metric, we then develop the task relation layer, which can be combined with any deep learning architecture to form task relation networks to fully exploit the relations of different tasks in an online fashion. Benefiting from the task relation layer, the task relation networks can better leverage the mutual information from the data. We demonstrate our proposed task relation networks are effective in improving the performance in both homogeneous and heterogeneous multi-task learning settings through extensive experiments on computer vision tasks.

An end-to-end network architecture, the Recurrent Shape Regression (RSR), is presented to deal with the task of facial shape detection, a crucial step in many computer vision problems. The RSR generalizes the conventional cascaded regression into a recurrent dynamic network through abstracting common latent models with stage-to-stage operations. Instead of invariant regression transformation, we construct shape-dependent dynamic regressors to attain the recurrence of regression action itself. The regressors can be stacked into a high-order regression network to represent more complex shape regression. By further integrating feature learning as well as global shape constraint, the RSR becomes more controllable in entire optimization of shape regression, where the gradient computation can be efficiently back-propagated through time. To handle the possible partial occlusions of shapes, we propose a mimic virtual occlusion strategy by randomly disturbing certain point cliques without the requirement of any annotations of occlusion information or even occluded training data. Extensive experiments on five face datasets demonstrate that the proposed RSR outperforms the recent state-of-the-art cascaded approaches.

This paper studies the subspace clustering problem. Given some data points approximately drawn from a union of subspaces, the goal is to group these data points into their underlying subspaces. Many subspace clustering methods have been proposed and among which sparse subspace clustering and low-rank representation are two representative ones. Despite the different motivations, we observe that many existing methods own the common block diagonal property, which possibly leads to correct clustering, yet with their proofs given case by case. In this work, we consider a general formulation and provide a unified theoretical guarantee of the block diagonal property. The block diagonal property of many existing methods falls into our special case. Second, we observe that many existing methods approximate the block diagonal representation matrix by using different structure priors, e.g., sparsity and low-rankness, which are indirect. We propose the first block diagonal matrix induced regularizer for directly pursuing the block diagonal matrix. With this regularizer, we solve the subspace clustering problem by Block Diagonal Representation (BDR), which uses the block diagonal structure prior. The BDR model is nonconvex and we propose an alternating minimization solver and prove its convergence. Experiments on real datasets demonstrate the effectiveness of BDR.

Modeling the aging process of human faces is important for cross-age face verification and recognition. In this paper, we propose a Recurrent Face Aging (RFA) framework which takes as input a single image and automatically outputs a series of aged faces. The hidden units in the RFA are connected autoregressively allowing the framework to age the person by referring to the previous aged faces. Due to the lack of labeled face data of the same person captured in a long range of ages, traditional face aging models split the ages into discrete groups and learn a one-step face transformation for each pair of adjacent age groups. Since human face aging is a smooth progression, it is more appropriate to age the face by going through smooth transitional states. In this way, the intermediate aged faces between the age groups can be generated. Towards this target, we employ a recurrent neural network whose recurrent module is a hierarchical triple-layer gated recurrent unit which functions as an autoencoder. The bottom layer of the module encodes the input to a latent representation, and the top layer decodes the representation to a corresponding aged face. The experimental results demonstrate the effectiveness of our framework.

Synthesizing realistic profile faces is beneficial for more efficiently training deep pose-invariant models for large-scale unconstrained face recognition, by augmenting the number of samples with extreme poses and avoiding costly annotation work. However, learning from synthetic faces may not achieve the desired performance due to the discrepancy betwedistributions of the synthetic and real face images. To narrow this gap, we propose a Dual-Agent Generative Adversarial Network (DA-GAN) model, which can improve the realism of a face simulator's output using unlabeled real faces while preserving the identity information during the realism refinement. The dual agents are specially designed for distinguishing real versus fake and identities simultaneously. In particular, we employ an off-the-shelf 3D face model as a simulator to generate profile face images with varying poses. DA-GAN leverages a fully convolutional network as the generator to generate high-resolution images and an auto-encoder as the discriminator with the dual agents. Besides the novel architecture, we make several key modifications to the standard GAN to preserve pose, texture as well as identity, and stabilize the training process: (i) a pose perception loss; (ii) an identity perception loss; (iii) an adversarial loss with a boundary equilibrium regularization term. Experimental results show that DA-GAN not only achieves outstanding perceptual results but also significantly outperforms state-of-the-arts on the large-scale and challenging NIST IJB-A and CFP unconstrained face recognition benchmarks. In addition, the proposed DA-GAN is also a promising new approach for solving generic transfer learning problems more effectively. DA-GAN is the foundation of our winning entry to the NIST IJB-A face recognition competition in which we secured the $1^{st}$ places on the tracks of verification and identification.

Kernel methods have been successfully applied to the areas of pattern recognition and data mining. In this paper, we mainly discuss the issue of propagating labels in kernel space. A Kernel-Induced Label Propagation (Kernel-LP) framework by mapping is proposed for high-dimensional data classification using the most informative patterns of data in kernel space. The essence of Kernel-LP is to perform joint label propagation and adaptive weight learning in a transformed kernel space. That is, our Kernel-LP changes the task of label propagation from the commonly-used Euclidean space in most existing work to kernel space. The motivation of our Kernel-LP to propagate labels and learn the adaptive weights jointly by the assumption of an inner product space of inputs, i.e., the original linearly inseparable inputs may be mapped to be separable in kernel space. Kernel-LP is based on existing positive and negative LP model, i.e., the effects of negative label information are integrated to improve the label prediction power. Also, Kernel-LP performs adaptive weight construction over the same kernel space, so it can avoid the tricky process of choosing the optimal neighborhood size suffered in traditional criteria. Two novel and efficient out-of-sample approaches for our Kernel-LP to involve new test data are also presented, i.e., (1) direct kernel mapping and (2) kernel mapping-induced label reconstruction, both of which purely depend on the kernel matrix between training set and testing set. Owing to the kernel trick, our algorithms will be applicable to handle the high-dimensional real data. Extensive results on real datasets demonstrate the effectiveness of our approach.

In this paper, we aim to build a comprehensive face detection system which provides a one-stop solution to various practical challenges for face detection in realistic scenarios, e.g., detecting faces from multiple-views, faces with occlusions, exaggerated expressions or blurred faces. Moreover, we introduce an automatic data harvest algorithm to effectively improve the generalization performance of the system even when collecting training faces containing various challenging patterns is difficult. In particular, we introduce three critical components to build the system, i.e., a recently widely used deep convolutional neural network (CNN), a novel blur-aware bi-channel network architecture, and a new self-learning mechanism capable of exploiting video contexts continuously. The aforementioned challenges except for detecting blurred faces can potentially be addressed by the CNN component owing its robustness to local deformation of target faces. The more challenging problem of detecting blurred faces is addressed by the bi-channel architecture component which processes blurred and clear faces adaptively. In addition, to address the difficulties in improving the generalization performance of the learning-based face detection system, we introduce a video-context-based self-learning mechanism into the system, which enables the system to continuously enhance its performance by harvesting faces with challenging training patterns automatically. To exploit video context, the detector is applied to massive unlabeled videos, and challenging faces are captured based on temporal inference. These recaptured faces, generally corresponding to one or multiple challenges mentioned above, are fed into the detection system to further improve its performance. Extensive experiments with the proposed detection system provide new state-of-the-art performance on FDDB data set, PASCAL face data set, AFW data set, and WIDER Face data set.

In this paper, we revisit the classical stochastic jump-diffusion process and develop an effective variant for estimating visibility statuses of objects while tracking them in videos. Dealing with partial or full occlusions is a long standing problem in computer vision but largely remains unsolved. In this paper, we cast the above problem as a Markov decision process and develop a policy-based jump-diffusion method to jointly track object locations in videos and estimate their visibility statuses. Our method employs a set of jump dynamics to change visibility statuses of objects and a set of diffusion dynamics to track objects in videos. Different from the traditional jump-diffusion process that stochastically generates dynamics, we utilize deep policy functions to determine the best dynamic for the present state and learn the optimal policies using reinforcement learning methods. Our method is capable of tracking objects with full or partial occlusions in crowded scenes. We evaluate the proposed method over challenging video sequences and compare it to alternative tracking methods. Significant improvements are made particularly for videos with frequent interactions or occlusions.

Unpaired image-to-image translation problem aims to model the mapping from one domain to another with unpaired training data. Current works like the well-acknowledged Cycle GAN provide a general solution for any two domains through modeling injective mappings with a symmetric structure. While in situations where two domains are asymmetric in complexity, i.e., the amount of information between two domains is different, these approaches pose problems of poor generation quality, mapping ambiguity, and model sensitivity. To address these issues, we propose Asymmetric GAN (AsymGAN) to adapt the asymmetric domains by introducing an auxiliary variable (aux) to learn the extra information for transferring from the information-poor domain to the information-rich domain, which improves the performance of state-of-the-art approaches in the following ways. First, aux better balances the information between two domains which benefits the quality of generation. Second, the imbalance of information commonly leads to mapping ambiguity, where we are able to model one-to-many mappings by tuning aux, and furthermore, our aux is controllable. Third, the training of Cycle GAN can easily make the generator pair sensitive to small disturbances and variations while our model decouples the ill-conditioned relevance of generators by injecting aux during training. We verify the effectiveness of our proposed method both qualitatively and quantitatively on asymmetric situation, label-photo task, on Cityscapes and Helen datasets, and show many applications of asymmetric image translations. In conclusion, our AsymGAN provides a better solution for unpaired image-to-image translation in asymmetric domains.

Predicting human pose in the wild is a challenging problem due to high flexibility of joints and possible occlusion. Existing approaches generally tackle the difficulties either by holistic prediction or multi-stage processing, which suffer from poor performance for locating challenging joints or high computational cost. In this paper, we propose a new Hierarchical Contextual Refinement Network (HCRN) to robustly predict human poses in an efficient manner, where human body joints of different complexities are processed at different layers in a context hierarchy. Different from existing approaches, our proposed model predicts positions of joints from easy to difficult in a single stage through effectively exploiting informative contexts provided in the previous layer. Such approach offers two appealing advantages over state-of-the-arts: (1) more accurate than predicting all the joints together and (2) more efficient than multi-stage processing methods. We design a Contextual Refinement Unit (CRU) to implement the proposed model, which enables auto-diffusion of joint detection results to effectively transfer informative context from easy joints to difficult ones. In this way, difficult joints can be reliably detected even in presence of occlusion or severe distracting factors. Multiple CRUs are organized into a tree-structured hierarchy which is end-to-end trainable and does not require processing joints for multiple iterations. Comprehensive experiments evaluate the efficacy and efficiency of the proposed HCRN model to improve well-established baselines and achieve new state-of-the-art on multiple human pose estimation benchmarks.

In this paper, we propose a novel unsupervised Nonnegative Adaptive Feature Extraction (NAFE) algorithm for data representation and classification. The formulation of NAFE integrates the sparsity constrained nonnegative matrix factorization (NMF), representation learning, and adaptive reconstruction weight learning into a unified model. Specifically, NAFE performs feature and weight learning over the new robust representations of NMF for more accurate measure and representation. For nonnegative adaptive feature extraction, our NAFE first utilizes the sparsity constrained NMF to obtain the new and robust representations of the original data. To preserve the manifold structures of the learnt new representations, we also incorporate a neighborhood reconstruction error over the weight matrix for joint minimization. Note that to further improve the representation power, the weights are jointly shared in the new low-dimensional nonnegative representation space, low-dimensional nonlinear manifold space, and low-dimensional projective subspace, i.e., local neighborhood information is clearly preserved in different feature spaces so that informative representations and features can be jointly obtained. To enable NAFE to extract features from new data, we also include a feature approximation error by a linear projection so that the learnt extractor can obtain features from new data efficiently. Extensive simulations show that our formulation can deliver state-of-the-art results on several public databases for feature extraction and classification, compared with several related methods.

Tensor data (i.e., the data having multiple dimensions) are quickly growing in scale in many practical applications, which poses new challenges for data modeling and analysis approaches, such as high-order relations of large complexity, gross noise, and varying data scale. Existing low-rank data analysis methods, which are effective at analyzing matrix data, may fail in the regime of tensor data due to these challenges. A robust and scalable low-rank tensor modeling method is heavily desired. In this paper, we develop an online robust low-rank tensor modeling (ORLTM) method to address these challenges. The ORLTM method leverages the high-order correlations among all tensor modes to model an intrinsic low-rank structure of streaming tensor data online and can effectively analyze data residing in a mixture of multiple subspaces by virtue of dictionary learning. ORLTM consumes a very limited memory space that remains constant regardless of the increase of tensor data size, which facilitates processing tensor data at a large scale. More concretely, it models each mode unfolding of streaming tensor data using the bilinear formulation of tensor nuclear norms. With this reformulation, ORLTM employs a stochastic optimization algorithm to learn the tensor low-rank structure alternatively for online updating. To capture the final tensors, ORLTM uses an average pooling operation on folded tensors in all modes. We also provide the analysis regarding computational complexity, memory cost, and convergence. Moreover, we extend ORLTM to the image alignment scenario by incorporating the geometrical transformations and linearizing the constraints. Extensive empirical studies on synthetic database and three practical vision tasks, including video background subtraction, image alignment, and visual tracking, have demonstrated the superiority of the proposed method.

Spectral Clustering (SC) is a widely used data clustering method which first learns a low-dimensional embedding U of data by computing the eigenvectors of the normalized Laplacian matrix, and then performs k-means on UT to get the final clustering result. The Sparse Spectral Clustering (SSC) method extends SC with a sparse regularization on UUT by using the block diagonal structure prior of UUT in the ideal case. However, encouraging UUT to be sparse leads to a heavily nonconvex problem which is challenging to solve and the work (Lu, Yan, and Lin 2016) proposes a convex relaxation in the pursuit of this aim indirectly. However, the convex relaxation generally leads to a loose approximation and the quality of the solution is not clear. This work instead considers to solve the nonconvex formulation of SSC which directly encourages UUT to be sparse. We propose an efficient Alternating Direction Method of Multipliers (ADMM) to solve the nonconvex SSC and provide the convergence guarantee. In particular, we prove that the sequences generated by ADMM always exist a limit point and any limit point is a stationary point. Our analysis does not impose any assumptions on the iterates and thus is practical. Our proposed ADMM for nonconvex problems allows the stepsize to be increasing but upper bounded, and this makes it very efficient in practice. Experimental analysis on several real data sets verifies the effectiveness of our method.

Previous deep learning based state-of-the-art scene text detection methods can be roughly classified into two categories. The first category treats scene text as a type of general objects and follows general object detection paradigm to localize scene text by regressing the text box locations, but troubled by the arbitrary-orientation and large aspect ratios of scene text. The second one segments text regions directly, but mostly needs complex post processing. In this paper, we present a method that combines the ideas of the two types of methods while avoiding their shortcomings. We propose to detect scene text by localizing corner points of text bounding boxes and segmenting text regions in relative positions. In inference stage, candidate boxes are generated by sampling and grouping corner points, which are further scored by segmentation maps and suppressed by NMS. Compared with previous methods, our method can handle long oriented text naturally and doesn't need complex post processing. The experiments on ICDAR2013, ICDAR2015, MSRA-TD500, MLT and COCO-Text demonstrate that the proposed algorithm achieves better or comparable results in both accuracy and efficiency. Based on VGG16, it achieves an F-measure of 84.3% on ICDAR2015 and 81.5% on MSRA-TD500.

Human pose estimation still faces various difficulties in challenging scenarios. Human parsing, as a closely related task, can provide valuable cues for better pose estimation, which however has not been fully exploited. In this paper, we propose a novel Parsing Induced Learner to exploit parsing information to effectively assist pose estimation by learning to fast adapt the base pose estimation model. The proposed Parsing Induced Learner is composed of a parsing encoder and a pose model parameter adapter, which together learn to predict dynamic parameters of the pose model to extract complementary useful features for more accurate pose estimation. Comprehensive experiments on benchmarks LIP and extended PASCAL-Person-Part show that the proposed Parsing Induced Learner can improve performance of both single- and multi-person pose estimation to new state-of-the-art. Cross-dataset experiments also show that the proposed Parsing Induced Learner from LIP dataset can accelerate learning of a human pose estimation model on MPII benchmark in addition to achieving outperforming performance.

In this paper we propose a noval method to generate the specified network parameters through one feed-forward propagation in the meta networks for neural style transfer. Recent works on style transfer typically need to train image transformation networks for every new style, and the style is encoded in the network parameters by enormous iterations of stochastic gradient descent, which lacks the generalization ability to new style in the inference stage. To tackle these issues, we build a meta network which takes in the style image and generates a corresponding image transformation network directly. Compared with optimization-based methods for every style, our meta networks can handle an arbitrary new style within 19 milliseconds on one modern GPU card. The fast image transformation network generated by our meta network is only 449 KB, which is capable of real-time running on a mobile device. We also investigate the manifold of the style transfer networks by operating the hidden features from meta networks. Experiments have well validated the effectiveness of our method. Code and trained models will be released.

Pose variation is one key challenge in face recognition. As opposed to current techniques for pose invariant face recognition, which either directly extract pose invariant features for recognition, or first normalize profile face images to frontal pose before feature extraction, we argue that it is more desirable to perform both tasks jointly to allow them to benefit from each other. To this end, we propose a Pose Invariant Model (PIM) for face recognition in the wild, with three distinct novelties. First, PIM is a novel and unified deep architecture, containing a Face Frontalization sub-Net (FFN) and a Discriminative Learning sub-Net (DLN), which are jointly learned from end to end. Second, FFN is a well-designed dual-path Generative Adversarial Network (GAN) which simultaneously perceives global structures and local details, incorporated with an unsupervised cross-domain adversarial training and a "learning to learn" strategy for high-fidelity and identity-preserving frontal view synthesis. Third, DLN is a generic Convolutional Neural Network (CNN) for face recognition with our enforced cross-entropy optimization strategy for learning discriminative yet generalized feature representation. Qualitative and quantitative experiments on both controlled and in-the-wild benchmarks demonstrate the superiority of the proposed model over the state-of-the-arts.

We present a new approach and a novel architecture, termed WSNet, for learning compact and efficient deep neural networks. Existing approaches conventionally learn full model parameters independently and then compress them via ad hoc processing such as model pruning or filter factorization. Alternatively, WSNet proposes learning model parameters by sampling from a compact set of learnable parameters, which naturally enforces parameter sharing throughout the learning process. We demonstrate that such a novel weight sampling approach (and induced WSNet) promotes both weights and computation sharing favorably. By employing this method, we can more efficiently learn much smaller networks with competitive performance compared to baseline networks with equal numbers of convolution filters. Specifically, we consider learning compact and efficient 1D convolutional neural networks for audio classification. Extensive experiments on multiple audio classification datasets verify the effectiveness of WSNet. Combined with weight quantization, the resulted models are up to 180x smaller and theoretically up to 16x faster than the well-established baselines, without noticeable performance drop.

In this paper, we discuss the robust factorization based robust dictionary learning problem for data representation. A Robust Projective Low-Rank and Sparse Representation model (R-PLSR) is technically proposed. Our R-PLSR model integrates the L1-norm based robust factorization and robust low-rank & sparse representation by robust dictionary learning into a unified framework. Specifically, R-PLSR performs the joint low-rank and sparse representation over the informative low-dimensional representations by robust sparse factorization so that the results are more accurate. To make the factorization and representation procedures robust to noise and outliers, R-PLSR imposes the sparse L2, 1-norm jointly on the reconstruction errors based on the factorization and dictionary learning. Note that L2, 1-norm can also minimize the reconstruction error as much as possible, since the L2, 1-norm theoretically tends to force many rows of the reconstruction error matrix to be zeros. The Nuclear-norm and L1-norm are jointly used on the representation coefficients so that salient representations can be obtained. Extensive results on several image datasets show that our R-PLSR formulation can deliver superior performance over other state-of-the-arts.

In this paper, we mainly propose a Robust Adaptive Projective Dictionary Pair Learning (RA-DPL) framework based on the adaptive discriminative representations. Our formulation can seamlessly integrate the robust projective dictionary pair learning and the adaptive sparse representation learning into a unified model. RA-DPL improves the existing DPL algorithm in threefold. First, RA-DPL aims at computing the robust projective dictionary pairs by employing the sparse and robust l2,1-norm to encode the reconstruction error. Second, RA-DPL regularizes the robust l2,1-norm on the analysis dictionary so that the analysis dictionary can extract sparse coefficients from the given samples explicitly. More importantly, the optimization of l2,1-norm is so efficient, that is, the sparse coding step will be time-saving. Third, RA-DPL can clearly preserve the local neighborhood relationship of the sparse coefficients within each class, which can make the learnt representations discriminating and can also improve the discriminating power of learnt dictionary. Extensive simulations on image databases demonstrate that our RA-DPL can obtain the superior performance over other state-of-the-arts.

We mainly propose a robust Embedded Locality-Constrained Label Consistent Dictionary Learning (ELC2DL) framework for discriminative classification. ELC2DL improves the representation and classification performance by performing DL in the noise-removed sparse embedding space, since most real data often contains noise and performing DL over noisy data for reconstruction may decrease performance potentially. To reduce the noise in data, our model computes a sparse projection jointly for noise reduction and then uses the noise-removed data for DL. By incorporating a noise-reduction term with a discriminative locality-constrained label consistent term that associates the label information with each dictionary atom to preserve local structure of training data, a noise-reduction projection, an over-complete dictionary and discriminative sparse codes are obtained jointly. Simulations on several image databases show that our algorithm can deliver enhanced performance over other state-of-the-arts.

The residual unit and its variations are wildly used in building very deep neural networks for alleviating optimization difficulty. In this work, we revisit the standard residual function as well as its several successful variants and propose a unified framework based on tensor Block Term Decomposition (BTD) to explain these apparently different residual functions from the tensor decomposition view. With the BTD framework, we further propose a novel basic network architecture, named the Collective Residual Unit (CRU). CRU further enhances parameter efficiency of deep residual neural networks by sharing core factors derived from collective tensor factorization over the involved residual units. It enables efficient knowledge sharing across multiple residual units, reduces the number of model parameters, lowers the risk of over-fitting, and provides better generalization ability. Extensive experimental results show that our proposed CRU network brings outstanding parameter efficiency -- it achieves comparable classification performance with ResNet-200 while using a model size as small as ResNet-50 on the ImageNet-1k and Places365-Standard benchmark datasets.

The recent proposed Tensor Nuclear Norm (TNN) [Lu et al., 2016; 2018a] is an interesting convex penalty induced by the tensor SVD [Kilmer and Martin, 2011]. It plays a similar role as the matrix nuclear norm which is the convex surrogate of the matrix rank. Considering that the TNN based Tensor Robust PCA [Lu et al., 2018a] is an elegant extension of Robust PCA with a similar tight recovery bound, it is natural to solve other low rank tensor recovery problems extended from the matrix cases. However, the extensions and proofs are generally tedious. The general atomic norm provides a unified view of low-complexity structures induced norms, e.g., the l1-norm and nuclear norm. The sharp estimates of the required number of generic measurements for exact recovery based on the atomic norm are known in the literature. In this work, with a careful choice of the atomic set, we prove that TNN is a special atomic norm. Then by computing the Gaussian width of certain cone which is necessary for the sharp estimate, we achieve a simple bound for guaranteed low tubal rank tensor recovery from Gaussian measurements. Specifically, we show that by solving a TNN minimization problem, the underlying tensor of size n1×n2×n3 with tubal rank r can be exactly recovered when the given number of Gaussian measurements is O(r(n1+n2−r)n3). It is order optimal when comparing with the degrees of freedom r(n1+n2−r)n3. Beyond the Gaussian mapping, we also give the recovery guarantee of tensor completion based on the uniform random mapping by TNN minimization. Numerical experiments verify our theoretical results.

Both accuracy and speed are equally important in urban scene parsing. Most of the existing methods mainly focus on improving parsing accuracy, ignoring the problem of low inference speed due to large-sized input and high resolution feature maps. To tackle this issue, we propose a High Resolution Feature Recovering (HRFR) framework to accelerate a given parsing network. A Super-Resolution Recovering module is employed to recover features of large original-sized images from features of down-sampled input. Therefore, our framework can combine the advantages of (1) fast speed of networks with down-sampled input and (2) high accuracy of networks with large original-sized input. Additionally, we employ auxiliary intermediate supervision and boundary region re-weighting to facilitate the optimization of the network. Extensive experiments on the two challenging Cityscapes and CamVid datasets well demonstrate the effectiveness of the proposed HRFR framework, which can accelerate the scene parsing inference process by about 3.0x speedup from 1/2 down-sampled input with negligible accuracy reduction.

Learning from synthetic faces, though perhaps appealing for high data efficiency, may not bring satisfactory performance due to the distribution discrepancy of the synthetic and real face images. To mitigate this gap, we propose a 3D-Aided Deep Pose-Invariant Face Recognition Model (3D-PIM), which automatically recovers realistic frontal faces from arbitrary poses through a 3D face model in a novel way. Specifically, 3D-PIM incorporates a simulator with the aid of a 3D Morphable Model (3D MM) to obtain shape and appearance prior for accelerating face normalization learning, requiring less training data. It further leverages a global-local Generative Adversarial Network (GAN) with multiple critical improvements as a refiner to enhance the realism of both global structures and local details of the face simulator’s output using unlabelled real data only, while preserving the identity information. Qualitative and quantitative experiments on both controlled and in-the-wild benchmarks clearly demonstrate superiority of the proposed model over state-of-the-arts.

Human parsing is an important task in human-centric analysis. Despite the remarkable progress in single-human parsing, the more realistic case of multi-human parsing remains challenging in terms of the data and the model. Compared with the considerable number of available single-human parsing datasets, the datasets for multi-human parsing are very limited in number mainly due to the huge annotation effort required. Besides the data challenge to multi-human parsing, the persons in real-world scenarios are often entangled with each other due to close interaction and body occlusion, making it difficult to distinguish body parts from different person instances. In this paper we propose the Multi-Human Parsing Machines (MHPM) system, which contains an MHP Montage model and an MHP Solver, to address both challenges in multi-human parsing. Specifically, the MHP Montage model in MHPM generates realistic images with multiple persons together with the parsing labels. It intelligently composes single persons onto background scene images while maintaining the structural information between persons and the scene. The generated images can be used to train better multi-human parsing algorithms. On the other hand, the MHP Solver in MHPM solves the bottleneck of distinguishing multiple entangled persons with close interaction. It employs a Group-Individual Push and Pull (GIPP) loss function, which can effectively separate persons with close interaction. We experimentally show that the proposed MHPM can achieve state-of-the-art performance on the multi-human parsing benchmark and the person individualization benchmark, which distinguishes closely entangled person instances.

Style synthesis attracts great interests recently, while few works focus on its dual problem "style separation". In this paper, we propose the Style Separation and Synthesis Generative Adversarial Network (S3-GAN) to simultaneously implement style separation and style synthesis on object photographs of specific categories. Based on the assumption that the object photographs lie on a manifold, and the contents and styles are independent, we employ S3-GAN to build mappings between the manifold and a latent vector space for separating and synthesizing the contents and styles. The S3-GAN consists of an encoder network, a generator network, and an adversarial network. The encoder network performs style separation by mapping an object photograph to a latent vector. Two halves of the latent vector represent the content and style, respectively. The generator network performs style synthesis by taking a concatenated vector as input. The concatenated vector contains the style half vector of the style target image and the content half vector of the content target image. Once obtaining the images from the generator network, an adversarial network is imposed to generate more photo-realistic images. Experiments on CelebA and UT Zappos 50K datasets demonstrate that the S3-GAN has the capacity of style separation and synthesis simultaneously, and could capture various styles in a single model.

Despite the noticeable progress in perceptual tasks like detection, instance segmentation and human parsing, computers still perform unsatisfactorily on visually understanding humans in crowded scenes, such as group behavior analysis, person re-identification and autonomous driving, etc. To this end, models need to comprehensively perceive the semantic information and the differences between instances in a multi-human image, which is recently defined as the multi-human parsing task. In this paper, we present a new large-scale database "Multi-Human Parsing (MHP)" for algorithm development and evaluation, and advances the state-of-the-art in understanding humans in crowded scenes. MHP contains 25,403 elaborately annotated images with 58 fine-grained semantic category labels, involving 2-26 persons per image and captured in real-world scenes from various viewpoints, poses, occlusion, interactions and background. We further propose a novel deep Nested Adversarial Network (NAN) model for multi-human parsing. NAN consists of three Generative Adversarial Network (GAN)-like sub-nets, respectively performing semantic saliency prediction, instance-agnostic parsing and instance-aware clustering. These sub-nets form a nested structure and are carefully designed to learn jointly in an end-to-end way. NAN consistently outperforms existing state-of-the-art solutions on our MHP and several other datasets, and serves as a strong baseline to drive the future research for multi-human parsing.

Learning to capture long-range relations is fundamental to image/video recognition. Existing CNN models generally rely on increasing depth to model such relations which is highly inefficient. In this work, we propose the “double attention block”, a novel component that aggregates and propagates informative global features from the entire spatio-temporal space of input images/videos, enabling subsequent convolution layers to access features from the entire space efficiently. The component is designed with a double attention mechanism in two steps, where the first step gathers features from the entire space into a compact set through second-order attention pooling and the second step adaptively selects and distributes features to each location via another attention. The proposed double attention block is easy to adopt and can be plugged into existing deep neural networks conveniently. We conduct extensive ablation studies and experiments on both image and video recognition tasks for evaluating its performance. On the image recognition task, a ResNet-50 equipped with our double attention blocks outperforms a much larger ResNet-152 architecture on ImageNet-1k dataset with over 40% less the number of parameters and less FLOPs. On the action recognition task, our proposed model achieves the state-of-the-art results on the Kinetics and UCF-101 datasets with significantly higher efficiency than recent works.

Instance-level object segmentation is an important yet under-explored task. Most of state-of-the-art methods rely on region proposal methods to extract candidate segments and then utilize object classification to produce final results. Nonetheless, generating reliable region proposals itself is a quite challenging and unsolved task. In this work, we propose a Proposal-Free Network (PFN) to address the instance-level object segmentation problem, which outputs the numbers of instances of different categories and the pixel-level information on i) the coordinates of the instance bounding box each pixel belongs to, and ii) the confidences of different categories for each pixel, based on pixel-to-pixel deep convolutional neural network. All the outputs together, by using any off-the-shelf clustering method for simple post-processing, can naturally generate the ultimate instance-level object segmentation results. The whole PFN can be easily trained without the requirement of a proposal generation stage. Extensive evaluations on the challenging PASCAL VOC 2012 semantic segmentation benchmark demonstrate the effectiveness of the proposed PFN solution without relying on any proposal generation methods.

Accompanied with the rising popularity of compressed sensing, the Alternating Direction Method of Multipliers (ADMM) has become the most widely used solver for linearly constrained convex problems with separable objectives. In this work, we observe that many existing ADMMs update the primal variable by minimizing different majorant functions with their convergence proofs given case by case. Inspired by the principle of majorization minimization, we respectively present the unified frameworks of Gauss-Seidel ADMMs and Jacobian ADMMs, which use different historical information for the current updating. Our frameworks generalize previous ADMMs to solve the problems with non-separable objectives. We also show that ADMMs converge faster when the used majorant function is tighter. We then propose the Mixed Gauss-Seidel and Jacobian ADMM (M-ADMM) which alleviates the slow convergence issue of Jacobian ADMMs by absorbing merits of the Gauss-Seidel ADMMs. M-ADMM can be further improved by backtracking and wise variable partition. We also propose to solve the multi-blocks problems by Proximal Gauss-Seidel ADMM which is of the Gauss-Seidel type. It convegences for non-strongly convex objective. Experiments on both synthesized and real-world data demonstrate the superiority of our new ADMMs. Finally, we release a toolbox that implements efficient ADMMs for many problems in compressed sensing.

Age progression is defined as aesthetically re-rendering the aging face at any future age for an individual face. In this work, we aim to automatically render aging faces in a personalized way. Basically, for each age group, we learn an aging dictionary to reveal its aging characteristics (e.g., wrinkles), where the dictionary bases corresponding to the same index yet from two neighboring aging dictionaries form a particular aging pattern cross these two age groups, and a linear combination of all these patterns expresses a particular personalized aging process. Moreover, two factors are taken into consideration in the dictionary learning process. First, beyond the aging dictionaries, each person may have extra personalized facial characteristics, e.g., mole, which are invariant in the aging process. Second, it is challenging or even impossible to collect faces of all age groups for a particular person, yet much easier and more practical to get face pairs from neighboring age groups. To this end, we propose a novel Bi-level Dictionary Learning based Personalized Age Progression (BDL-PAP) method. Here, bi-level dictionary learning is formulated to learn the aging dictionaries based on face pairs from neighboring age groups. Extensive experiments well demonstrate the advantages of the proposed BDL-PAP over other state-of-the-arts in term of personalized age progression, as well as the performance gain for cross-age face verification by synthesizing aging faces.

Face alignment acts as an important task in computer vision. Regression-based methods currently dominate the approach to solving this problem, which generally employ a series of mapping functions from the face appearance to iteratively update the face shape hypothesis. One keypoint here is thus how to perform the regression procedure. In this work, we formulate this regression procedure as a sparse coding problem. We learn two relational dictionaries, one for the face appearance and the other one for the face shape, with coupled reconstruction coefficient to capture their underlying relationships. To deploy this model for face alignment, we derive the relational dictionaries in a stage-wised manner to perform close-loop refinement of themselves, i.e., the face appearance dictionary is first learned from the face shape dictionary and then used to update the face shape hypothesis, and the updated face shape dictionary from the shape hypothesis is in return used to refine the face appearance dictionary. To improve the model accuracy, we extend this model hierarchically from the whole face shape to face part shapes, thus both the global and local view variations of a face are captured. To locate facial landmarks under occlusions, we further introduce an occlusion dictionary into the face appearance dictionary to recover face shape from partially occluded face appearance. The occlusion dictionary is learned in a data driven manner from background images to represent a set of elemental occlusion patterns, a sparse combination of which models various practical partial face occlusions. By integrating all these technical innovations, we obtain a robust and accurate approach to locate facial landmarks under different face views and possibly severe occlusions for face images in the wild. Extensive experimental analyses and evaluations on different benchmark datasets, as well as two new datasets built by ourselves, have demonstrated the robustness and accuracy of our proposed model, especially for face images with large view variations and/or severe occlusions.

Object proposal generation, as a preprocessing technique, has been widely used in current object detection pipelines to guide the search of objects and avoid exhaustive sliding window search across images. Current object proposals are mostly based on low-level image cues, such as edges and saliency. However, objectness is possibly a high-level semantic concept showing whether one region contains objects. This paper presents a framework utilizing fully convolutional networks (FCNs) to produce object proposal positions and bounding box location refinement with Support Vector Machine (SVM) to further improve proposal localization. Experiments on the PASCAL VOC 2007 show that using high-level semantic object proposals obtained by FCN, the object recall can be improved. An improvement in detection mean average precision is also seen when using our proposals in the Fast R-convolutional neural network framework. In addition, we also demonstrate that our method shows stronger robustness when introduced to image perturbations, e.g., blurring, JPEG compression, and salt and pepper noise. Finally, the generalization capability of our model (trained on the PASCAL VOC 2007) is evaluated and validated by testing on PASCAL VOC 2012 validation set, ILSVRC 2013 validation set, and MS COCO 2014 validation set.

We propose a novel end-to-end deep architecture for face landmark detection, based on a deep convolutional and deconvolutional network followed by carefully designed recurrent network structures. The pipeline of this architecture consists of three parts. Through the first part, we encode an input face image to resolution-preserved deconvolutional feature maps via a deep network with stacked convolutional and deconvolutional layers. Then, in the second part, we estimate the initial coordinates of the facial key points by an additional convolutional layer on top of these deconvolutional feature maps. In the last part, by using the deconvolutional feature maps and the initial facial key points as input, we refine the coordinates of the facial key points by a recurrent network that consists of multiple long short-term memory components. Extensive evaluations on several benchmark data sets show that the proposed deep architecture has superior performance against the state-of-the-art methods.

Video-based person re-identification plays a central role in realistic security and video surveillance. In this paper, we propose a novel accumulative motion context (AMOC) network for addressing this important problem, which effectively exploits the long-range motion context for robustly identifying the same person under challenging conditions. Given a video sequence of the same or different persons, the proposed AMOC network jointly learns appearance representation and motion context from a collection of adjacent frames using a two-stream convolutional architecture. Then, AMOC accumulates clues from motion context by recurrent aggregation, allowing effective information flow among adjacent frames and capturing dynamic gist of the persons. The architecture of AMOC is end-to-end trainable, and thus, motion context can be adapted to complement appearance clues under unfavorable conditions (e.g., occlusions). Extensive experiments are conduced on three public benchmark data sets, i.e., the iLIDS-VID, PRID-2011, and MARS data sets, to investigate the performance of AMOC. The experimental results demonstrate that the proposed AMOC network outperforms state-of-the-arts for video-based re-identification significantly and confirm the advantage of exploiting long-range motion context for video-based person re-identification, validating our motivation evidently.

In this paper, we propose a novel fine-grained dictionary learning method for image classification. To learn a high-quality discriminative dictionary, three types of multispecific subdictionaries, i.e., class-specific dictionaries (CSDs), universal dictionary (UD), and family-specific dictionaries (FSDs), are simultaneously uncovered. Here, CSDs and UD, respectively, model the patterns for each class and the patterns irrespective of any class. FSDs can help reveal the shared patterns between multiple image classes, by filling the gap between the patterns in CSDs and UD. The dependence among image classes is revealed by the shared FSDs, and a common FSD can be assigned to several classes to represent their residual. Finally, the most discriminative FSD for each class is identified by minimizing the sparse reconstruction error. Extensive experiments are conducted on different widely used data sets for image classification. The results demonstrate the superior performance of the proposed method over some state-of-the-art methods.

Most previous works on the first-person video recognition focus on measuring the similarity of different actions by using low-level features of objects interacting with humans. However, due to noisy camera motion and frequent changes in viewpoint and scale, they fail to capture and model highly discriminative object features. In this paper, we propose a novel pipeline for the first-person daily activity recognition. Our object feature extraction pipeline is inspired by the recent success of object hypotheses and deep convolutional neural network (CNN)-based detection frameworks. Our key contribution is a simple yet effective manipulated object proposal generation scheme. This scheme leverages motion cues, such as motion boundary and motion magnitude (in contrast, camera motion is usually considered as “noise” for most previous methods), to generate a more compact and discriminative set of object proposals, which are more closely related to the objects, which are being manipulated. Then, we learn more discriminative object detectors from these manipulated object proposals based on region-based CNN. Meanwhile, we develop a non-linear feature fusion scheme, which better combines object and motion features. We show in experiments that the proposed framework significantly outperforms the state-of-the-art recognition performance on a challenging first-person daily activity benchmark.

Coherent image regions can be used as good features for many computer vision tasks, such as object tracking, segmentation, and recognition. Most of previous region extraction methods, however, are not suitable for online applications because of their either heavy computations or unsatisfactory results. We propose a seed-based region growing and merging approach to generate simultaneously coherent and discriminative image regions. We present a quadtree-based seed initialization algorithm to adaptively place seeds into different image areas and then grow them into regions by a color- and edge-guided growing procedure. To merge these regions in different levels, we propose to use the generalized boundary strength to measure the quality of region merging result. In addition, we present a region merging algorithm of linear time complexity to perform efficient and effective region merging. Overall, our new approach simultaneously holds these advantages: 1) it is extremely fast with linear complexity in both time and space, which takes less than 50 ms to process an HVGA image; 2) it can give a direct control of the region number and well adapt to image regions with various sizes and shapes; and 3) it provides a tree-structured representation of the regions and thus can model the image from multiple scales. We evaluate the proposed approach on the standard benchmarks with extensive comparisons with the state-of-the-art methods. The experimental results demonstrate its good comprehensive performances. Example applications using the extracted regions as features for online object tracking and multiclass object segmentation also exhibit its potential for many computer vision tasks.

In this paper, we focus on improving the proposal classification stage in the object detection task and present implicit negative sub-categorization and sink diversion to lift the performance by strengthening loss function in this stage. First, based on the observation that the "background" class is generally very diverse and thus challenging to be handled as a single indiscriminative class in existing state-of-the-art methods, we propose to divide the background category into multiple implicit sub-categories to explicitly differentiate diverse patterns within it. Second, since the ground truth class inevitably has low-value probability scores for certain images, we propose to add a "sink" class and divert the probabilities of wrong classes to this class when necessary, such that the ground truth label will still have a higher probability than other wrong classes even though it has low probability output. Additionally, we propose to use dilated convolution, which is widely used in the semantic segmentation task, for efficient and valuable context information extraction. Extensive experiments on PASCAL VOC 2007 and 2012 data sets show that our proposed methods based on faster R-CNN implementation can achieve state-of-the-art mAPs, i.e., 84.1%, 82.6%, respectively, and obtain 2.5% improvement on ILSVRC DET compared with that of ResNet.

Face attribute prediction in the wild is important for many facial analysis applications yet it is very challenging due to ubiquitous face variations. In this paper, we address face attribute prediction in the wild by proposing a novel method, lAndmark Free Face AttrIbute pRediction (AFFAIR). Unlike traditional face attribute prediction methods that require facial landmark detection and face alignment, AFFAIR uses an endto- end learning pipeline to jointly learn a hierarchy of spatial transformations that optimize facial attribute prediction with no reliance on landmark annotations or pre-trained landmark detectors. AFFAIR achieves this through simultaneously 1) learning a global transformation which effectively alleviates negative effect of global face variation for the following attribute prediction tailored for each face, 2) locating the most relevant facial part for attribute prediction and 3) aggregating the global and local features for robust attribute prediction. Within AFFAIR, a new competitive learning strategy is developed that effectively enhances global transformation learning for better attribute prediction. We show that with zero information about landmarks, AFFAIR achieves state-of-the-art performance on three face attribute prediction benchmarks, which simultaneously learns the face-level transformation and attribute-level localization within a unified framework.

Graph-based dimensionality reduction techniques have been widely and successfully applied to clustering and classification tasks. The basis of these algorithms is the constructed graph which dictates their performance. In general, the graph is defined by the input affinity matrix. However, the affinity matrix derived from the data is sometimes suboptimal for dimension reduction as the data used are very noisy. To address this issue, we propose the projective unsupervised flexible embedding models with optimal graph (PUFE-OG). We build an optimal graph by adjusting the affinity matrix. To tackle the out-of-sample problem, we employ a linear regression term to learn a projection matrix. The optimal graph and the projection matrix are jointly learned by integrating the manifold regularizer and regression residual into a unified model. The experimental results on the public benchmark datasets demonstrate that the proposed PUFE-OG outperforms state-of-the-art methods.

Face recognition techniques have been developed significantly in recent years. However, recognizing faces with partial occlusion is still challenging for existing face recognizers, which is heavily desired in real-world applications concerning surveillance and security. Although much research effort has been devoted to developing face de-occlusion methods, most of them can only work well under constrained conditions, such as all of faces are from a pre-defined closed set of subjects. In this paper, we propose a robust LSTM-Autoencoders (RLA) model to effectively restore partially occluded faces even in the wild. The RLA model consists of two LSTM components, which aims at occlusion-robust face encoding and recurrent occlusion removal respectively. The first one, named multi-scale spatial LSTM encoder, reads facial patches of various scales sequentially to output a latent representation, and occlusion-robustness is achieved owing to the fact that the influence of occlusion is only upon some of the patches. Receiving the representation learned by the encoder, the LSTM decoder with a dual channel architecture reconstructs the overall face and detects occlusion simultaneously, and by feat of LSTM, the decoder breaks down the task of face de-occlusion into restoring the occluded part step by step. Moreover, to minimize identify information loss and guarantee face recognition accuracy over recovered faces, we introduce an identity-preserving adversarial training scheme to further improve RLA. Extensive experiments on both synthetic and real data sets of faces with occlusion clearly demonstrate the effectiveness of our proposed RLA in removing different types of facial occlusion at various locations. The proposed method also provides significantly larger performance gain than other de-occlusion methods in promoting recognition performance over partially-occluded faces.

This article presents a high-precision multi-modal approach for localizing moving cameras with monocular videos, which has wide potentials in many intelligent applications, including robotics, autonomous vehicles, and so on. Existing visual odometry methods often suffer from symmetric or repetitive scene patterns, e.g., windows on buildings or parking stalls. To address this issue, we introduce a robust camera localization method that contributes in two aspects. First, we formulate feature tracking, the critical step of visual odometry, as a hierarchical min-cost network flow optimization task, and we regularize the formula with flow constraints, cross-scale consistencies, and motion heuristics. The proposed regularized formula is capable of adaptively selecting distinctive features or feature combinations, which is more effective than traditional methods that detect and group repetitive patterns in a separate step. Second, we develop a joint formula for integrating dense visual odometry and sparse GPS readings in a common reference coordinate. The fusion process is guided with high-order statistics knowledge to suppress the impacts of noises, clusters, and model drifting. We evaluate the proposed camera localization method on both public video datasets and a newly created dataset that includes scenes full of repetitive patterns. Results with comparisons show that our method can achieve comparable performance to state-of-the-art methods and is particularly effective for addressing repetitive pattern issues.

This paper presents a weakly supervised regression model for the visual house appraisal problem, which aims to predict the value of a house from its photos and textual descriptions (e.g., number of bedrooms). The key idea of our approach is a multi-layer neural network, called multi-instance Deep Ranking and Regression (MiDRR) net, which jointly solves two coupled tasks: ranking and regression, in the multiple instance setting. The network is trained using weakly supervised data, which do not require intensive human annotations. We also design a set of human heuristics to promote deep features through imposing constraints over the solution space, e.g., a house with three bedrooms often has a higher value than that with only two bedrooms. While these constraints are specific to the studied problem, the developed formula can be easily generalized to the other regression applications. For test and evaluation purposes, we collect a comprehensive house image benchmark that includes 900,000 photos from 30,000 houses recently traded in the USA, and apply the proposed MiDRR net to predict house values. Extensive evaluations with comparisons demonstrate that additional usage of imagery data as well as human heuristics can significantly boost system performance and that the proposed MiDRR net clearly outperforms the alternative methods.

Most existing detection pipelines treat object proposals independently and predict bounding box locations and classification scores over them separately. However, the important semantic and spatial layout correlations among proposals are often ignored, which are actually useful for more accurate object detection. In this paper, we propose a new EM-like group recursive learning approach to iteratively refine object proposals by incorporating such context of surrounding proposals and provide an optimal spatial configuration of object detections. In addition, we propose to incorporate the weakly supervised object segmentation cues and region-based object detection into a multistage architecture in order to fully exploit the learned segmentation features for better object detection in an end-toend way. The proposed architecture consists of three cascaded networks that, respectively, learn to perform weakly supervised object segmentation, object proposal generation, and recursive detection refinement. Combining the group recursive learning and the multistage architecture provides competitive mAPs of 78.7% and 74.9% on the PASCAL VOC2007 and VOC2012 datasets, respectively, which outperform many well-established baselines significantly.

In this paper, we consider the problem of pedestrian detection in natural scenes. Intuitively, instances of pedestrians with different spatial scales may exhibit dramatically different features. Thus, large variance in instance scales, which results in undesirable large intracategory variance in features, may severely hurt the performance of modern object instance detection methods. We argue that this issue can be substantially alleviated by the divide-and-conquer philosophy. Taking pedestrian detection as an example, we illustrate how we can leverage this philosophy to develop a Scale-Aware Fast R-CNN (SAF R-CNN) framework. The model introduces multiple built-in subnetworks which detect pedestrians with scales from disjoint ranges. Outputs from all of the subnetworks are then adaptively combined to generate the final detection results that are shown to be robust to large variance in instance scales, via a gate function defined over the sizes of object proposals. Extensive evaluations on several challenging pedestrian detection datasets well demonstrate the effectiveness of the proposed SAF R-CNN. Particularly, our method achieves state-of-the-art performance on Caltech, and obtains competitive results on INRIA, ETH, and KITTI.

In this paper, we propose two novel components for improving deep salient object detection models. The first component, called saliency detection network (S-Net), introduces dense short- and long-range connections that effectively integrate multiscale features to better exploit contexts at multiple levels. Benefiting from the direct access to low- and high-level features, the S-Net can not only exploit the object context but also preserve the object boundary sharply, leading to enhanced saliency detection performance. Second, a distraction detection network (D-Net) is developed to learn to diagnose which regions of an input image are distracting and harmful for saliency prediction of the S-Net. With such distraction diagnosis, the regions that are distracting to S-Net are removed in hindsight from the input image and the resulted distraction-free image is fed to S-Net for saliency prediction. To train the D-Net, a distraction mining approach is proposed to localize the model-specific distracting regions through examining the sensitiveness of the S-Net to image regions in a principled manner. Besides, the distraction mining approach also provides a way to interpret decisions made by deep neural network (DNN) saliency detection models, which relieves the black-box issues of DNNs to some extent. Extensive experiments on seven popular benchmark datasets demonstrate the effectiveness of the combined S-Net and D-Net, which provides new state of the arts.

Deep learning (DL) methods and architectures have been the state-of-the-art classification algorithms for computer vision and natural language processing problems. However, the successful application of these methods in motor imagery (MI) brain-computer interfaces (BCIs), in order to boost classification performance, is still limited. In this paper, we propose a classification framework for MI data by introducing a new temporal representation of the data and also utilizing a convolutional neural network (CNN) architecture for classification. The new representation is generated from modifying the filter-bank common spatial patterns method, and the CNN is designed and optimized accordingly for the representation. Our framework outperforms the best classification method in the literature on the BCI competition IV-2a 4-class MI data set by 7% increase in average subject accuracy. Furthermore, by studying the convolutional weights of the trained networks, we gain an insight into the temporal characteristics of EEG.

In this paper, we propose an analysis mechanism-based structured analysis discriminative dictionary learning analysis discriminative dictionary learning, framework. The ADDL seamlessly integrates ADDL, analysis representation, and analysis classifier training into a unified model. The applied analysis mechanism can make sure that the learned dictionaries, representations, and linear classifiers over different classes are independent and discriminating as much as possible. The dictionary is obtained by minimizing a reconstruction error and an analytical incoherence promoting term that encourages the subdictionaries associated with different classes to be independent. To obtain the representation coefficients, ADDL imposes a sparse -norm constraint on the coding coefficients instead of using or norm, since the - or -norm constraint applied in most existing DL criteria makes the training phase time consuming. The code-extraction projection that bridges data with the sparse codes by extracting special features from the given samples is calculated via minimizing a sparse code approximation term. Then we compute a linear classifier based on the approximated sparse codes by an analysis mechanism to simultaneously consider the classification and representation powers. Thus, the classification approach of our model is very efficient, because it can avoid the extra time-consuming sparse reconstruction process with trained dictionary for each new test data as most existing DL algorithms. Simulations on real image databases demonstrate that our ADDL model can obtain superior performance over other state of the arts.

We propose a robust inductive semi-supervised label prediction model over the embedded representation, termed adaptive embedded label propagation with weight learning (AELP-WL), for classification. AELP-WL offers several properties. First, our method seamlessly integrates the robust adaptive embedded label propagation with adaptive weight learning into a unified framework. By minimizing the reconstruction errors over embedded features and embedded soft labels jointly, our AELP-WL can explicitly ensure the learned weights to be joint optimal for representation and classification, which differs from most existing LP models that perform weight learning separately by an independent step before label prediction. Second, existing models usually precalculate the weights over the original samples that may contain unfavorable features and noise decreasing performance. To this end, our model adds a constraint that decomposes original data into a sparse component encoding embedded noise-removed sparse representations of samples and a sparse error part fitting noise, and then performs the adaptive weight learning over the embedded sparse representations. Third, our AELP-WL computes the projected soft labels by trading-off the manifold smoothness and label fitness errors over the adaptive weights and the embedded representations for enhancing the label estimation power. By including a regressive label approximation error for simultaneous minimization to correlate sample features with the embedded soft labels, the out-of-sample issue is naturally solved. By minimizing the reconstruction errors over features and embedded soft labels, classification error and label approximation error jointly, state-of-the-art results are delivered.

In this paper, we consider the scene parsing problem and propose a novel Multi-Path Feedback recurrent neural network (MPF-RNN) for parsing scene images. MPF-RNN can enhance the capability of RNNs in modeling long-range context information at multiple levels and better distinguish pixels that are easy to confuse. Different from feedforward CNNs and RNNs with only single feedback, MPF-RNN propagates the contextual features learned at top layer through multiple weighted recurrent connections to learn bottom features. For better training MPF-RNN, we propose a new strategy that considers accumulative loss at multiple recurrent steps to improve performance of the MPF-RNN on parsing small objects. With these two novel components, MPF-RNN has achieved significant improvement over strong baselines (VGG16 and Res101) on five challenging scene parsing benchmarks, including traditional SiftFlow, Barcelona, CamVid, Stanford Background as well as the recently released large-scale ADE20K.

In this paper, we present a novel and general network structure towards accelerating the inference process of convolutional neural networks, which is more complicated in network structure yet with less inference complexity. The core idea is to equip each original convolutional layer with another low-cost collaborative layer (LCCL), and the element-wise multiplication of the ReLU outputs of these two parallel layers produces the layer-wise output. The combined layer is potentially more discriminative than the original convolutional layer, and its inference is faster for two reasons: 1) the zero cells of the LCCL feature maps will remain zero after element-wise multiplication, and thus it is safe to skip the calculation of the corresponding high-cost convolution in the original convolutional layer; 2) LCCL is very fast if it is implemented as a 1 × 1 convolution or only a single filter shared by all channels. Extensive experiments on the CIFAR-10, CIFAR-100 and ILSCRC-2012 benchmarks show that our proposed network structure can accelerate the inference process by 32% on average with negligible performance drop.

Detecting small objects is notoriously challenging due to their low resolution and noisy representation. Existing object detection pipelines usually detect small objects through learning representations of all the objects at multiple scales. However, the performance gain of such ad hoc architectures is usually limited to pay off the computational cost. In this work, we address the small object detection problem by developing a single architecture that internally lifts representations of small objects to super-resolved ones, achieving similar characteristics as large objects and thus more discriminative for detection. For this purpose, we propose a new Perceptual Generative Adversarial Network (Perceptual GAN) model that improves small object detection through narrowing representation difference of small objects from the large ones. Specifically, its generator learns to transfer perceived poor representations of the small objects to super-resolved ones that are similar enough to real large objects to fool a competing discriminator. Meanwhile its discriminator competes with the generator to identify the generated representation and imposes an additional perceptual requirement - generated representations of small objects must be beneficial for detection purpose - on the generator. Extensive evaluations on the challenging Tsinghua-Tencent 100K [45] and the Caltech [9] benchmark well demonstrate the superiority of Perceptual GAN in detecting small objects, including traffic signs and pedestrians, over well-established state-of-the-arts.

This paper develops a general framework for learning interpretable data representation via Long Short-Term Memory (LSTM) recurrent neural networks over hierarchal graph structures. Instead of learning LSTM models over the pre-fixed structures, we propose to further learn the intermediate interpretable multi-level graph structures in a progressive and stochastic way from data during the LSTM network optimization. We thus call this model the structure-evolving LSTM. In particular, starting with an initial element-level graph representation where each node is a small data element, the structure-evolving LSTM gradually evolves the multi-level graph representations by stochastically merging the graph nodes with high compatibilities along the stacked LSTM layers. In each LSTM layer, we estimate the compatibility of two connected nodes from their corresponding LSTM gate outputs, which is used to generate a merging probability. The candidate graph structures are accordingly generated where the nodes are grouped into cliques with their merging probabilities. We then produce the new graph structure with a Metropolis-Hasting algorithm, which alleviates the risk of getting stuck in local optimums by stochastic sampling with an acceptance probability. Once a graph structure is accepted, a higher-level graph is then constructed by taking the partitioned cliques as its nodes. During the evolving process, representation becomes more abstracted in higher-levels where redundant information is filtered out, allowing more efficient propagation of long-range data dependencies. We evaluate the effectiveness of structure-evolving LSTM in the application of semantic object parsing and demonstrate its advantage over state-of-the-art LSTM models on standard benchmarks.

This paper describes a fast and accurate semantic image segmentation approach that encodes not only segmentation-specified features but also high-order context compatibilities and boundary guidance constraints. We introduce a structured patch prediction technique to make a trade-off between classification discriminability and boundary sensibility for features. Both label and feature contexts are embedded to ensure recognition accuracy and compatibility, while the complexity of the high order cliques is reduced by a distance-aware sampling and pooling strategy. The proposed joint model also employs a guidance CRF to further enhance the segmentation performance. The message passing step is augmented with the guided filtering which enables an efficient and joint training of the whole system in an end-to-end fashion. Our proposed joint model outperforms the state-of-art on Pascal VOC 2012 and Cityscapes, with mIoU(%) of 82.5 and 79.2 respectively. It also reaches a leading performance on ADE20K, which is the dataset of the scene parsing track in ILSVRC 2016. The code is available at https://github.com/FalongShen/SegModel.

We investigate a principle way to progressively mine discriminative object regions using classification networks to address the weakly-supervised semantic segmentation problems. Classification networks are only responsive to small and sparse discriminative regions from the object of interest, which deviates from the requirement of the segmentation task that needs to localize dense, interior and integral regions for pixel-wise inference. To mitigate this gap, we propose a new adversarial erasing approach for localizing and expanding object regions progressively. Starting with a single small object region, our proposed approach drives the classification network to sequentially discover new and complement object regions by erasing the current mined regions in an adversarial manner. These localized regions eventually constitute a dense and complete object region for learning semantic segmentation. To further enhance the quality of the discovered regions by adversarial erasing, an online prohibitive segmentation learning approach is developed to collaborate with adversarial erasing by providing auxiliary segmentation supervision modulated by the more reliable classification scores. Despite its apparent simplicity, the proposed approach achieves 55.0% and 55.7% mean Intersection-over-Union (mIoU) scores on PASCAL VOC 2012 val and test sets, which are the new state-of-the-arts.

In this paper, we address a rain removal problem from a single image, even in the presence of heavy rain and rain streak accumulation. Our core ideas lie in our new rain image model and new deep learning architecture. We add a binary map that provides rain streak locations to an existing model, which comprises a rain streak layer and a background layer. We create a model consisting of a component representing rain streak accumulation (where individual streaks cannot be seen, and thus visually similar to mist or fog), and another component representing various shapes and directions of overlapping rain streaks, which usually happen in heavy rain. Based on the model, we develop a multi-task deep learning architecture that learns the binary rain streak map, the appearance of rain streaks, and the clean background, which is our ultimate output. The additional binary map is critically beneficial, since its loss function can provide additional strong information to the network. To handle rain streak accumulation (again, a phenomenon visually similar to mist or fog) and various shapes and directions of overlapping rain streaks, we propose a recurrent rain detection and removal network that removes rain streaks and clears up the rain accumulation iteratively and progressively. In each recurrence of our method, a new contextualized dilated network is developed to exploit regional contextual information and to produce better representations for rain detection. The evaluation on real images, particularly on heavy rain, shows the effectiveness of our models and architecture.

We introduce a new fashion search protocol where attribute manipulation is allowed within the interaction between users and search engines, e.g. manipulating the color attribute of the clothing from red to blue. It is particularly useful for image-based search when the query image cannot perfectly match users expectation of the desired product. To build such a search engine, we propose a novel memory-augmented Attribute Manipulation Network (AMNet) which can manipulate image representation at the attribute level. Given a query image and some attributes that need to modify, AMNet can manipulate the intermediate representation encoding the unwanted attributes and change them to the desired ones through following four novel components: (1) a dual-path CNN architecture for discriminative deep attribute representation learning, (2) a memory block with an internal memory and a neural controller for prototype attribute representation learning and hosting, (3) an attribute manipulation network to modify the representation of the query image with the prototype feature retrieved from the memory block, (4) a loss layer which jointly optimizes the attribute classification loss and a triplet ranking loss over triplet images for facilitating precise attribute manipulation and image retrieving. Extensive experiments conducted on two large-scale fashion search datasets, i.e. DARN and DeepFashion, have demonstrated that AMNet is able to achieve remarkably good performance compared with well-designed baselines in terms of effectiveness of attribute manipulation and search accuracy.

Video scene parsing is challenging due to the following two reasons: firstly, it is non-trivial to learn meaningful video representations for producing the temporally consistent labeling map; secondly, such a learning process becomes more difficult with insufficient labeled video training data. In this work, we propose a unified framework to address the above two problems, which is to our knowledge the first model to employ predictive feature learning in the video scene parsing. The predictive feature learning is carried out in two predictive tasks: frame prediction and predictive parsing. It is experimentally proved that the learned predictive features in our model are able to significantly enhance the video parsing performance by combining with the standard image parsing network. Interestingly, the performance gain brought by the predictive learning is almost costless as the features are learned from a large amount of unlabeled video data in an unsupervised way. Extensive experiments over two challenging datasets, Cityscapes and Camvid, have demonstrated the effectiveness of our model by showing remarkable improvement over well-established baselines.

Parsing urban scene images benefits many applications, especially self-driving. Most of the current solutions employ generic image parsing models that treat all scales and locations in the images equally and do not consider the geometry property of car-captured urban scene images. Thus, they suffer from heterogeneous object scales caused by perspective projection of cameras on actual scenes and inevitably encounter parsing failures on distant objects as well as other boundary and recognition errors. In this work, we propose a new FoveaNet model to fully exploit the perspective geometry of scene images and address the common failures of generic parsing models. FoveaNet estimates the perspective geometry of a scene image through a convolutional network which integrates supportive evidence from contextual objects within the image. Based on the perspective geometry information, FoveaNet “undoes” the camera perspective projection - analyzing regions in the space of the actual scene, and thus provides much more reliable parsing results. Furthermore, to effectively address the recognition errors, FoveaNet introduces a new dense CRFs model that takes the perspective geometry as a prior potential. We evaluate FoveaNet on two urban scene parsing datasets, Cityspaces and CamVid, which demonstrates that FoveaNet can outperform all the well-established baselines and provide new state-of-the-art performance.

We investigate the problem of person search in the wild in this work. Instead of comparing the query against all candidate regions generated in a query-blind manner, we propose to recursively shrink the search area from the whole image till achieving precise localization of the target person, by fully exploiting information from the query and contextual cues in every recursive search step. We develop the Neural Person Search Machines (NPSM) to implement such recursive localization for person search. Benefiting from its neural search mechanism, NPSM is able to selectively shrink its focus from a loose region to a tighter one containing the target automatically. In this process, NPSM employs an internal primitive memory component to memorize the query representation which modulates the attention and augments its robustness to other distracting regions. Evaluations on two benchmark datasets, CUHK-SYSU Person Search dataset and PRW dataset, have demonstrated that our method can outperform current state-of-the-arts in both mAP and top-1 evaluation protocols.

Despite remarkable progress of face analysis techniques, detecting landmarks on large-pose faces is still difficult due to self-occlusion, subtle landmark difference and incomplete information. To address these challenging issues, we introduce a novel recurrent 3D-2D dual learning model that alternatively performs 2D-based 3D face model refinement and 3D-to-2D projection based 2D landmark refinement to reliably reason about self-occluded landmarks, precisely capture the subtle landmark displacement and accurately detect landmarks even in presence of extremely large poses. The proposed model presents the first loop-closed learning framework that effectively exploits the informative feedback from the 3D-2D learning and its dual 2D-3D refinement tasks in a recurrent manner. Benefiting from these two mutual-boosting steps, our proposed model demonstrates appealing robustness to large poses (up to profile pose) and outstanding ability to capture fine-scale landmark displacement compared with existing 3D models. It achieves new state-of-the-art on the challenging AFLW benchmark. Moreover, our proposed model introduces a new architectural design that economically utilizes intermediate features and achieves 4× faster speed than its deep learning based counterparts.

Many existing scene parsing methods adopt Convolutional Neural Networks with fixed-size receptive fields, which frequently result in inconsistent predictions of large objects and invisibility of small objects. To tackle this issue, we propose a scale-adaptive convolution to acquire flexiblesize receptive fields during scene parsing. Through adding a new scale regression layer, we can dynamically infer the position-adaptive scale coefficients which are adopted to resize the convolutional patches. Consequently, the receptive fields can be adjusted automatically according to the various sizes of the objects in scene images. Thus, the problems of invisible small objects and inconsistent large-object predictions can be alleviated. Furthermore, our proposed scale-adaptive convolutions are not only differentiable to learn the convolutional parameters and scale coefficients in an end-to-end way, but also of high parallelizability for the convenience of GPU implementation. Additionally, since the new scale regression layers are learned implicitly, any extra training supervision of object sizes is unnecessary. Extensive experiments on Cityscapes and ADE20K datasets well demonstrate the effectiveness of the proposed scaleadaptive convolutions.

Learning rich and diverse representations is critical for the performance of deep convolutional neural networks (CNNs). In this paper, we consider how to use privileged information to promote inherent diversity of a single CNN model such that the model can learn better representations and offer stronger generalization ability. To this end, we propose a novel group orthogonal convolutional neural network (GoCNN) that learns untangled representations within each layer by exploiting provided privileged information and enhances representation diversity effectively. We take image classification as an example where image segmentation annotations are used as privileged information during the training process. Experiments on two benchmark datasets – ImageNet and PASCAL VOC – clearly demonstrate the strong generalization ability of our proposed GoCNN model. On the ImageNet dataset, GoCNN improves the performance of state-of-the-art ResNet-152 model by absolute value of 1.2% while only uses privileged information of 10% of the training images, confirming effectiveness of GoCNN on utilizing available privileged knowledge to train better CNNs.

The rapid increase of multidimensional data (a.k.a. tensor) like videos brings new challenges for low-rank data modeling approaches such as dynamic data size, complex high-order relations, and multiplicity of low-rank structures. Resolving these challenges require a new tensor analysis method that can perform tensor data analysis online, which however is still absent. In this paper, we propose an Online Robust Low-rank Tensor Modeling (ORLTM) approach to address these challenges. ORLTM dynamically explores the high-order correlations across all tensor modes for low-rank structure modeling. To analyze mixture data from multiple subspaces, ORLTM introduces a new dictionary learning component. ORLTM processes data streamingly and thus requires quite low memory cost that is independent of data size. This makes ORLTM quite suitable for processing large-scale tensor data. Empirical studies have validated the effectiveness of the proposed method on both synthetic data and one practical task, i.e., video background subtraction. In addition, we provide theoretical analysis regarding computational complexity and memory cost, demonstrating the efficiency of ORLTM rigorously.

Most of existing scene parsing methods suffer from the serious problems of both inconsistent parsing results and object boundary shift. To tackle these problems, we first propose an iterative Global-residual Refinement Network (GRN) through exploiting global contextual information to predict the parsing residuals and iteratively smoothen the inconsistent parsing labels. Furthermore, we propose a Local-boundary Refinement Network (LRN) to learn the position-adaptive propagation coefficients so that local contextual information from neighbors can be optimally captured for refining object boundaries. Finally, we cascade the proposed two refinement networks after a fully residual convolutional neural network within a uniform framework. Extensive experiments on ADE20K and Cityscapes datasets well demonstrate the effectiveness of the two refinement methods for refining scene parsing predictions.

Face analytics benefits many multimedia applications. It consists of a number of tasks, such as facial emotion recognition and face parsing, and most existing approaches generally treat these tasks independently, which limits their deployment in real scenarios. In this paper we propose an integrated Face Analytics Network (iFAN), which is able to perform multiple tasks jointly for face analytics with a novel carefully designed network architecture to fully facilitate the informative interaction among different tasks. The proposed integrated network explicitly models the interactions between tasks so that the correlations between tasks can be fully exploited for performance boost. In addition, to solve the bottleneck of the absence of datasets with comprehensive training data for various tasks, we propose a novel cross-dataset hybrid training strategy. It allows "plug-in and play'' of multiple datasets annotated for different tasks without the requirement of a fully labeled common dataset for all the tasks. We experimentally show that the proposed iFAN achieves state-of-the-art performance on multiple face analytics tasks using a single integrated model. Specifically, iFAN achieves an overall F-score of 91.15% on the Helen dataset for face parsing, a normalized mean error of 5.81% on the MTFL dataset for facial landmark localization and an accuracy of 45.73% on the BNU dataset for emotion recognition with a single model.

Face aging, which renders aging faces for an input face, has attracted extensive attention in the multimedia research. Recently, several conditional Generative Adversarial Nets (GANs) based methods have achieved great success. They can generate images fitting the real face distributions conditioned on each individual age group. However, these methods fail to capture the transition patterns, e.g., the gradual shape and texture changes between adjacent age groups. In this paper, we propose a novel Contextual Generative Adversarial Nets (C-GANs) to specifically take it into consideration. The C-GANs consists of a conditional transformation network and two discriminative networks. The conditional transformation network imitates the aging procedure with several specially designed residual blocks. The age discriminative network guides the synthesized face to fit the real conditional distribution. The transition pattern discriminative network is novel, aiming to distinguish the real transition patterns with the fake ones. It serves as an extra regularization term for the conditional transformation network, ensuring the generated image pairs to fit the corresponding real transition pattern distribution. Experimental results demonstrate the proposed framework produces appealing results by comparing with the state-of-the-art and ground truth. We also observe performance gain for cross-age face verification.

Face aging, also known as age progression, is attracting more and more research interests. It has plenty of applications in various domains including cross-age face recognition, finding lost children, and entertainments. In recent years, face aging has witnessed various breakthroughs and a number of face aging models have been proposed. Face aging, however, is still a very challenging task in practice for various reasons. First, faces may have many different expressions and lighting conditions, which pose great challenges to modeling the aging patterns. Besides, the training data are usually very limited and the face images for the same person only cover a narrow range of ages.

In this work, we present a simple, highly efficient and modularized Dual Path Network (DPN) for image classification which presents a new topology of connection paths internally. By revealing the equivalence of the state-of-the-art Residual Network (ResNet) and Densely Convolutional Network (DenseNet) within the HORNN framework, we find that ResNet enables feature re-usage while DenseNet enables new features exploration which are both important for learning good representations. To enjoy the benefits from both path topologies, our proposed Dual Path Network shares common features while maintaining the flexibility to explore new features through dual path architectures. Extensive experiments on three benchmark datasets, ImagNet-1k, Places365 and PASCAL VOC, clearly demonstrate superior performance of the proposed DPN over state-of-the-arts. In particular, on the ImagNet-1k dataset, a shallow DPN surpasses the best ResNeXt-101(64x4d) with 26% smaller model size, 25% less computational cost and 8% lower memory consumption, and a deeper DPN (DPN-131) further pushes the state-of-the-art single model performance with about 2 times faster training speed. Experiments on the Places365 large-scale scene dataset, PASCAL VOC detection dataset, and PASCAL VOC segmentation dataset also demonstrate its consistently better performance than DenseNet, ResNet and the latest ResNeXt model over various applications.

It is important for intelligent systems, e.g. autonomous vehicles and robotics to anticipate the future in order to plan early and make decisions accordingly. Predicting the future scene parsing and motion dynamics helps the agents better understand the visual environment better as the former provides dense semantic segmentations, i.e. what objects will be present and where they will appear, while the latter provides dense motion information, i.e. how the objects move in the future. In this paper, we propose a novel model to predict the scene parsing and motion dynamics in unobserved future video frames simultaneously. Using history information (preceding frames and corresponding scene parsing results) as input, our model is able to predict the scene parsing and motion for arbitrary time steps ahead. More importantly, our model is superior compared to other methods that predict parsing and motion separately, as the complementary relationship between the two tasks are fully utilized in our model through joint learning. To our best knowledge, this is the first attempt in jointly predicting scene parsing and motion dynamics in the future frames. On the large-scale Cityscapes dataset, it is demonstrated that our model produces significantly better parsing and motion prediction results compared to well established baselines. In addition, we also show our model can be used to predict the steering angle of the vehicles, which further verifies the ability of our model to learn underlying latent parameters.

Synthesizing realistic profile faces is promising for more efficiently training deep pose-invariant models for large-scale unconstrained face recognition, by populating samples with extreme poses and avoiding tedious annotations. However, learning from synthetic faces may not achieve the desired performance due to the discrepancy between distributions of the synthetic and real face images. To narrow this gap, we propose a Dual-Agent Generative Adversarial Network (DA-GAN) model, which can improve the realism of a face simulator's output using unlabeled real faces, while preserving the identity information during the realism refinement. The dual agents are specifically designed for distinguishing real v.s. fake and identities simultaneously. In particular, we employ an off-the-shelf 3D face model as a simulator to generate profile face images with varying poses. DA-GAN leverages a fully convolutional network as the generator to generate high-resolution images and an auto-encoder as the discriminator with the dual agents. Besides the novel architecture, we make several key modifications to the standard GAN to preserve pose and texture, preserve identity and stabilize training process: (i) a pose perception loss; (ii) an identity perception loss; (iii) an adversarial loss with a boundary equilibrium regularization term. Experimental results show that DA-GAN not only presents compelling perceptual results but also significantly outperforms state-of-the-arts on the large-scale and challenging NIST IJB-A unconstrained face recognition benchmark. In addition, the proposed DA-GAN is also promising as a new approach for solving generic transfer learning problems more effectively.

An intuition on human segmentation is that when a human is moving in a video, the video-context (e.g., appearance and motion clues) may potentially infer reasonable mask information for the whole human body. Inspired by this, based on popular deep convolutional neural networks (CNN), we explore a very-weakly supervised learning framework for human segmentation task, where only an imperfect human detector is available along with massive weakly-labeled YouTube videos. In our solution, the video-context guided human mask inference and CNN based segmentation network learning iterate to mutually enhance each other until no further improvement gains. In the first step, each video is decomposed into supervoxels by the unsupervised video segmentation. The superpixels within the supervoxels are then classified as human or non-human by graph optimization with unary energies from the imperfect human detection results and the predicted confidence maps by the CNN trained in the previous iteration. In the second step, the video-context derived human masks are used as direct labels to train CNN. Extensive experiments on the challenging PASCAL VOC 2012 semantic segmentation benchmark demonstrate that the proposed framework has already achieved superior results than all previous weakly-supervised methods with object class or bounding box annotations. In addition, by augmenting with the annotated masks from PASCAL VOC 2012, our method reaches a new state-of-the-art performance on the human segmentation task.

In this work, we address the human parsing task with a novel Contextualized Convolutional Neural Network (Co-CNN) architecture, which well integrates the cross-layer context, global image-level context, semantic edge context, within-super-pixel context and cross-super-pixel neighborhood context into a unified network. Given an input human image, Co-CNN produces the pixelwise categorization in an end-to-end way. First, the cross-layer context is captured by our basic local-to-global-to-local structure, which hierarchically combines the global semantic information and the local fine details across different convolutional layers. Second, the global image-level label prediction is used as an auxiliary objective in the intermediate layer of the Co-CNN, and its outputs are further used for guiding the feature learning in subsequent convolutional layers to leverage the global image-level context. Third, semantic edge context is further incorporated into Co-CNN, where the high-level semantic boundaries are leveraged to guide pixel-wise labeling. Finally, to further utilize the local super-pixel contexts, the within-super-pixel smoothing and cross-super-pixel neighbourhood voting are formulated as natural sub-components of the Co-CNN to achieve the local label consistency in both training and testing process. Comprehensive evaluations on two public datasets well demonstrate the significant superiority of our Co-CNN over other state-of-the-arts for human parsing. In particular, the F-1 score on the large dataset [1] reaches 81.72 percent by Co-CNN, significantly higher than 62.81 percent and 64.38 percent by the state-of-the-art algorithms, M-CNN [2] and ATR [1], respectively. By utilizing our newly collected large dataset for training, our Co-CNN can achieve 85.36 percent in F-1 score.

Social image tag refinement, which aims to improve tag quality by automatically completing the missing tags and rectifying the noise-corrupted ones, is an essential component for social image search. Conventional approaches mainly focus on exploring the visual and tag information, without considering the user information, which often reveals important hints on the (in)correct tags of social images. Towards this end, we propose a novel tri-clustered tensor completion framework to collaboratively explore these three kinds of information to improve the performance of social image tag refinement. Specifically, the inter-relations among users, images and tags are modeled by a tensor, and the intra-relations between users, images and tags are explored by three regularizations respectively. To address the challenges of the super-sparse and large-scale tensor factorization that demands expensive computing and memory cost, we propose a novel tri-clustering method to divide the tensor into a certain number of sub-tensors by simultaneously clustering users, images and tags into a bunch of tri-clusters. And then we investigate two strategies to complete these sub-tensors by considering (in)dependence between the sub-tensors. Experimental results on a real-world social image database demonstrate the superiority of the proposed method compared with the state-of-the-art methods.

Recently, significant improvement has been made on semantic object segmentation due to the development of deep convolutional neural networks (DCNNs). Training such a DCNN usually relies on a large number of images with pixel-level segmentation masks, and annotating these images is very costly in terms of both finance and human effort. In this paper, we propose a simple to complex (STC) framework in which only image-level annotations are utilized to learn DCNNs for semantic segmentation. Specifically, we first train an initial segmentation network called Initial-DCNN with the saliency maps of simple images (i.e., those with a single category of major object(s) and clean background). These saliency maps can be automatically obtained by existing bottom-up salient object detection techniques, where no supervision information is needed. Then, a better network called Enhanced-DCNN is learned with supervision from the predicted segmentation masks of simple images based on the Initial-DCNN as well as the image-level annotations. Finally, more pixel-level segmentation masks of complex images (two or more categories of objects with cluttered background), which are inferred by using Enhanced-DCNN and image-level annotations, are utilized as the supervision information to learn the Powerful-DCNN for semantic segmentation. Our method utilizes 40K simple images from Flickr.com and 10K complex images from PASCAL VOC for step-wisely boosting the segmentation network. Extensive experimental results on PASCAL VOC 2012 segmentation benchmark well demonstrate the superiority of the proposed STC framework compared with other state-of-the-arts.

The human vision system usually has a specifically activated area of neurons when recognizing a category of images. Inspired by this visual mechanism, we propose a layerwise class-aware convolutional neural network architecture to explicitly discover category-tailored neurons on intermediate hidden layers to improve the network learning ability. Instead of directly selecting activated neurons for different categories, we inversely suppress those neurons of intermediate layers irrelevant with the given target class to produce a class-specific subnetwork, which implicitly enhances the discriminability of hidden layer features due to the increase of the inter-class discrepancy on them. Together with the classifier of the top layer, we jointly learn this network by formulating the suppressor of hidden layers as a penalty term in the objective function. To address class-specific neuron suppression in each hidden layer, we also introduce a statistic method based on mutual information to dynamically and automatically update the suppressed neurons during the network training. Extensive experiments demonstrate that the proposed model is superior to the state-of-the-art models.

Convolutional neural network (CNN) has achieved the state-of-the-art performance in many different visual tasks. Learned from a large-scale training data set, CNN features are much more discriminative and accurate than the handcrafted features. Moreover, CNN features are also transferable among different domains. On the other hand, traditional dictionary-based features (such as BoW and spatial pyramid matching) contain much more local discriminative and structural information, which is implicitly embedded in the images. To further improve the performance, in this paper, we propose to combine CNN with dictionary-based models for scene recognition and visual domain adaptation (DA). Specifically, based on the well-tuned CNN models (e.g., AlexNet and VGG Net), two dictionary-based representations are further constructed, namely, mid-level local representation (MLR) and convolutional Fisher vector (CFV) representation. In MLR, an efficient two-stage clustering method, i.e., weighted spatial and feature space spectral clustering on the parts of a single image followed by clustering all representative parts of all images, is used to generate a class-mixture or a class-specific part dictionary. After that, the part dictionary is used to operate with the multiscale image inputs for generating mid-level representation. In CFV, a multiscale and scale-proportional Gaussian mixture model training strategy is utilized to generate Fisher vectors based on the last convolutional layer of CNN. By integrating the complementary information of MLR, CFV, and the CNN features of the fully connected layer, the state-of-the-art performance can be achieved on scene recognition and DA problems. An interested finding is that our proposed hybrid representation (from VGG net trained on ImageNet) is also complementary to GoogLeNet and/or VGG-11 (trained on Place205) greatly.

This paper proposes a generic framework that enables a multiscale interaction in the cost aggregation step of stereo matching algorithms. Inspired by the formulation of image filters, we first reformulate cost aggregation from a weighted least-squares (WLS) optimization perspective and show that different cost aggregation methods essentially differ in the choices of similarity kernels. Our key motivation is that while the human stereo vision system processes information at both coarse and fine scales interactively for the correspondence search, state-of-the-art approaches aggregate costs at the finest scale of the input stereo images only, ignoring inter-consistency across multiple scales. This motivation leads us to introduce an inter-scale regularizer into the WLS optimization objective to enforce the consistency of the cost volume among the neighboring scales. The new optimization objective with the inter-scale regularization is convex, and thus, it is easily and analytically solved. Minimizing this new objective leads to the proposed framework. Since the regularization term is independent of the similarity kernel, various cost aggregation approaches, including discrete and continuous parameterization methods, can be easily integrated into the proposed framework. We show that the cross-scale framework is important as it effectively and efficiently expands state-of-the-art cost aggregation methods and leads to significant improvements, when evaluated on Middlebury, Middlebury Third, KITTI, and New Tsukuba data sets.

Recently, convolutional neural network (CNN) visual features have demonstrated their powerful ability as a universal representation for various recognition tasks. In this paper, cross-modal retrieval with CNN visual features is implemented with several classic methods. Specifically, off-the-shelf CNN visual features are extracted from the CNN model, which is pretrained on ImageNet with more than one million images from 1000 object categories, as a generic image representation to tackle cross-modal retrieval. To further enhance the representational ability of CNN visual features, based on the pretrained CNN model on ImageNet, a fine-tuning step is performed by using the open source Caffe CNN library for each target data set. Besides, we propose a deep semantic matching method to address the cross-modal retrieval problem with respect to samples which are annotated with one or multiple labels. Extensive experiments on five popular publicly available data sets well demonstrate the superiority of CNN visual features for cross-modal retrieval.

Age estimation based on the human face remains a significant problem in computer vision and pattern recognition. In order to estimate an accurate age or age group of a facial image, most of the existing algorithms require a huge face data set attached with age labels. This imposes a constraint on the utilization of the immensely unlabeled or weakly labeled training data, e.g., the huge amount of human photos in the social networks. These images may provide no age label, but it is easy to derive the age difference for an image pair of the same person. To improve the age estimation accuracy, we propose a novel learning scheme to take advantage of these weakly labeled data through the deep convolutional neural networks. For each image pair, Kullback-Leibler divergence is employed to embed the age difference information. The entropy loss and the cross entropy loss are adaptively applied on each image to make the distribution exhibit a single peak value. The combination of these losses is designed to drive the neural network to understand the age gradually from only the age difference information. We also contribute a data set, including more than 100 000 face images attached with their taken dates. Each image is both labeled with the timestamp and people identity. Experimental results on two aging face databases show the advantages of the proposed age difference learning system, and the state-of-the-art performance is gained.

Person re-identification across disjoint camera views has been widely applied in video surveillance yet it is still a challenging problem. One of the major challenges lies in the lack of spatial and temporal cues, which makes it difficult to deal with large variations of lighting conditions, viewing angles, body poses, and occlusions. Recently, several deep-learning-based person re-identification approaches have been proposed and achieved remarkable performance. However, most of those approaches extract discriminative features from the whole frame at one glimpse without differentiating various parts of the persons to identify. It is essentially important to examine multiple highly discriminative local regions of the person images in details through multiple glimpses for dealing with the large appearance variance. In this paper, we propose a new soft attention-based model, i.e., the end-to-end comparative attention network (CAN), specifically tailored for the task of person re-identification. The end-to-end CAN learns to selectively focus on parts of pairs of person images after taking a few glimpses of them and adaptively comparing their appearance. The CAN model is able to learn which parts of images are relevant for discerning persons and automatically integrates information from different parts to determine whether a pair of images belongs to the same person. In other words, our proposed CAN model simulates the human perception process to verify whether two images are from the same person. Extensive experiments on four benchmark person re-identification data sets, including CUHK01, CHUHK03, Market-1501, and VIPeR, clearly demonstrate that our proposed end-to-end CAN for person re-identification outperforms well established baselines significantly and offer the new state-of-the-art performance.

In this paper, we consider the image super-resolution (SR) problem. The main challenge of image SR is to recover high-frequency details of a low-resolution (LR) image that are important for human perception. To address this essentially ill-posed problem, we introduce a Deep Edge Guided REcurrent rEsidual (DEGREE) network to progressively recover the high-frequency details. Different from most of the existing methods that aim at predicting high-resolution (HR) images directly, the DEGREE investigates an alternative route to recover the difference between a pair of LR and HR images by recurrent residual learning. DEGREE further augments the SR process with edge-preserving capability, namely the LR image and its edge map can jointly infer the sharp edge details of the HR image during the recurrent recovery process. To speed up its training convergence rate, by-pass connections across the multiple layers of DEGREE are constructed. In addition, we offer an understanding on DEGREE from the view-point of sub-band frequency decomposition on image signal and experimentally demonstrate how the DEGREE can recover different frequency bands separately. Extensive experiments on three benchmark data sets clearly demonstrate the superiority of DEGREE over the well-established baselines and DEGREE also provides new state-of-the-arts on these data sets. We also present addition experiments for JPEG artifacts reduction to demonstrate the good generality and flexibility of our proposed DEGREE network to handle other image processing tasks.

We propose two nuclear- and L2,1-norm regularized 2D neighborhood preserving projection (2DNPP) methods for extracting representative 2D image features. 2DNPP extracts neighborhood preserving features by minimizing a Frobenius norm-based reconstruction error that is very sensitive noise and outliers in given data. To make the distance metric more reliable and robust, and encode the neighborhood reconstruction error more accurately, we minimize the nuclear- and L2,1-norm-based reconstruction error, respectively and measure it over each image. Technically, we propose two enhanced variants of 2DNPP, nuclear-norm-based 2DNPP and sparse reconstruction-based 2DNPP. Besides, to optimize the projection for more promising feature extraction, we also add the nuclear- and sparse L2,1-norm constraints on it accordingly, where L2,1-norm ensures the projection to be sparse in rows so that discriminative features are learnt in the latent subspace and the nuclear-norm ensures the low-rank property of features by projecting data into their respective subspaces. By fully considering the neighborhood preserving power, using more reliable and robust distance metric, and imposing the low-rank or sparse constraints on projections at the same time, our methods can outperform related state-of-the-arts in a variety of simulation settings.

Social media websites have become important information sharing platforms. The rapid development of social media platforms has led to increasingly large-scale social media data, which has shown remarkable societal and marketing values. There are needs to extract important events in live social media streams. However, microblogs event classification is challenging due to two facts, i.e., the short/conversational nature and the incompatible meanings between the text and the corresponding image in social posts, and the rapidly evolving contents. In this article, we propose to conduct event classification via deep learning and social tracking. First, we introduce a Multi-modal Multi-instance Deep Network (M 2 DN) for microblogs classification, which is able to handle the weakly labeled microblogs data oriented from the incompatible meanings inside microblogs. Besides predicting each microblogs as predefined events, we propose to employ social tracking to extract social-related auxiliary information to enrich the testing samples. We extract a set of candidate-relevant microblogs in a short time window by using social connections, such as related users and geographical locations. All these selected microblogs and the testing data are formulated in a Markov Random Field model. The inference on the Markov Random Field is conducted to update the classification results of the testing microblogs. This method is evaluated on the Brand-Social-Net dataset for classification of 20 events. Experimental results and comparison with the state of the arts show that the proposed method can achieve better performance for the event classification task.

Furniture style describes the discriminative appearance characteristics of furniture. It plays an important role in real-world indoor decoration. In this article, we explore the furniture style features and study the problem of furniture style classification. Differing from traditional object classification, furniture style classification aims at classifying different furniture in terms of the “style” that describes its appearance (e.g., American style, Gothic style, Rococo style, etc.) rather than the “kind” that is more related to its functional structure (e.g., bed, desk, etc.). To pursue efficient furniture style features, we construct a novel dataset of furniture styles that contains 16 common style categories and implement three strategies with respect to two categories of classification, that is, handcrafted classification and learning-based classification. First, we follow the typical image classification pipeline to extract the handcrafted features and train the classifier by support vector machine. Then we use the convolutional neural network to extract learning-based features from training images. To obtain comprehensive furniture style features, we finally combine the handcrafted image classification pipeline and the learning-based network. We experimentally evaluate the performances of handcrafted features and learning-based features of each strategy, and the results show the superiority of learning-based features and also the comprehensiveness of handcrafted features.

Modern deep neural network-based object detection methods typically classify candidate proposals using their interior features. However, global and local surrounding contexts that are believed to be valuable for object detection are not fully exploited by existing methods yet. In this work, we take a step towards understanding what is a robust practice to extract and utilize contextual information to facilitate object detection in practice. Specifically, we consider the following two questions: “how to identify useful global contextual information for detecting a certain object?” and “how to exploit local context surrounding a proposal for better inferring its contents?” We provide preliminary answers to these questions through developing a novel attention to context convolution neural network (AC-CNN)-based object detection model. AC-CNN effectively incorporates global and local contextual information into the region-based CNN (e.g., fast R-CNN and faster R-CNN) detection framework and provides better object detection performance. It consists of one attention-based global contextualized (AGC) subnetwork and one multi-scale local contextualized (MLC) subnetwork. To capture global context, the AGC subnetwork recurrently generates an attention map for an input image to highlight useful global contextual locations, through multiple stacked long short-term memory layers. For capturing surrounding local context, the MLC subnetwork exploits both the inside and outside contextual information of each specific proposal at multiple scales. The global and local context are then fused together for making the final decision for detection. Extensive experiments on PASCAL VOC 2007 and VOC 2012 well demonstrate the superiority of the proposed AC-CNN over well-established baselines.

Fine-grained object classification attracts increasing attention in multimedia applications. However, it is a quite challenging problem due to the subtle interclass difference and large intraclass variation. Recently, visual attention models have been applied to automatically localize the discriminative regions of an image for better capturing critical difference, which have demonstrated promising performance. Unfortunately, without consideration of the diversity in attention process, most of existing attention models perform poorly in classifying fine-grained objects. In this paper, we propose a diversified visual attention network (DVAN) to address the problem of fine-grained object classification, which substantially relieves the dependency on strongly supervised information for learning to localize discriminative regions compared with attention-less models. More importantly, DVAN explicitly pursues the diversity of attention and is able to gather discriminative information to the maximal extent. Multiple attention canvases are generated to extract convolutional features for attention. An LSTM recurrent unit is employed to learn the attentiveness and discrimination of attention canvases. The proposed DVAN has the ability to attend the object from coarse to fine granularity, and a dynamic internal representation for classification is built up by incrementally combining the information from different locations and scales of the image. Extensive experiments conducted on CUB-2011, Stanford Dogs, and Stanford Cars datasets have demonstrated that the pro-posed DVAN achieves competitive performance compared to the state-of-the-art approaches, without using any prior knowledge, user interaction, or external resource in training and testing.

Rectified linear activation units are important components for state-of-the-art deep convolutional networks. In this paper, we propose a novel S-shaped rectifiedlinear activation unit (SReLU) to learn both convexand non-convex functions, imitating the multiple function forms given by the two fundamental laws, namely the Webner-Fechner law and the Stevens law, in psychophysics and neural sciences. Specifically, SReLU consists of three piecewise linear functions, which are formulated by four learnable parameters. The SReLU is learned jointly with the training of the whole deep network through back propagation. During the training phase, to initialize SReLU in different layers, we propose a “freezing” method to degenerate SReLU into a predefined leaky rectified linear unit in the initial several training epochs and then adaptively learn the good initial values. SReLU can be universally used in the existing deep networks with negligible additional parameters and computation cost. Experiments with two popular CNN architectures, Network in Network and GoogLeNet on scale-various benchmarks including CIFAR10, CIFAR100, MNIST and ImageNet demonstrate that SReLU achieves remarkable improvement compared to other activation functions.

The Augmented Lagragian Method (ALM) and Alternating Direction Method of Multiplier (ADMM) have been powerful optimization methods for general convex programming subject to linear constraint. We consider the convex problem whose objective consists of a smooth part and a nonsmooth but simple part. We propose the Fast Proximal Augmented Lagragian Method (Fast PALM) which achieves the convergence rate O(1/K2), compared with O(1/K) by the traditional PALM. In order to further reduce the per-iteration complexity and handle the multi-blocks problem, we propose the Fast Proximal ADMM with Parallel Splitting (Fast PL-ADMM-PS) method. It also partially improves the rate related to the smooth part of the objective function. Experimental results on both synthesized and real world data demonstrate that our fast methods significantly improve the previous PALM and ADMM

Robust visual tracking is a challenging task in computer vision. Due to the accumulation and propagation of estimation error, model drifting often occurs and degrades the tracking performance. To mitigate this problem, in this paper we propose a novel tracking method called Recurrently Target-attending Tracking (RTT). RTT attempts to identify and exploit those reliable parts which are beneficial for the overall tracking process. To bypass occlusion and discover reliable components, multi-directional Recurrent Neural Networks (RNNs) are employed in RTT to capture long-range contextual cues by traversing a candidate spatial region from multiple directions. The produced confidence maps from the RNNs are employed to adaptively regularize the learning of discriminative correlation filters by suppressing clutter background noises while making full use of the information from reliable parts. To solve the weighted correlation filters, we especially derive an efficient closedform solution with a sharp reduction in computation complexity. Extensive experiments demonstrate that our proposed RTT is more competitive over those correlation filter based methods.

Semantic object parsing is a fundamental task for understanding objects in detail in computer vision community, where incorporating multi-level contextual information is critical for achieving such fine-grained pixel-level recognition. Prior methods often leverage the contextual information through post-processing predicted confidence maps. In this work, we propose a novel deep Local-Global Long Short-Term Memory (LG-LSTM) architecture to seamlessly incorporate short-distance and long-distance spatial dependencies into the feature learning over all pixel positions. In each LG-LSTM layer, local guidance from neighboring positions and global guidance from the whole image are imposed on each position to better exploit complex local and global contextual information. Individual LSTMs for distinct spatial dimensions are also utilized to intrinsically capture various spatial layouts of semantic parts in the images, yielding distinct hidden and memory cells of each position for each dimension. In our parsing approach, several LG-LSTM layers are stacked and appended to the intermediate convolutional layers to directly enhance visual features, allowing network parameters to be learned in an end-to-end way. The long chains of sequential computation by stacked LG-LSTM layers also enable each pixel to sense a much larger region for inference benefiting from the memorization of previous dependencies in all positions along all dimensions. Comprehensive evaluations on three public datasets well demonstrate the significant superiority of our LG-LSTM over other state-of-the-art methods.

In this work, we propose a novel Reversible Recursive Instance-level Object Segmentation (R2-IOS) framework to address the challenging instance-level object segmentation task. R2-IOS consists of a reversible proposal refinement sub-network that predicts bounding box offsets for refining the object proposal locations, and an instance-level segmentation sub-network that generates the foreground mask of the dominant object instance in each proposal. By being recursive, R2-IOS iteratively optimizes the two subnetworks during joint training, in which the refined object proposals and improved segmentation predictions are alternately fed into each other to progressively increase the network capabilities. By being reversible, the proposal refinement sub-network adaptively determines an optimal number of refinement iterations required for each proposal during both training and testing. Furthermore, to handle multiple overlapped instances within a proposal, an instance-aware denoising autoencoder is introduced into the segmentation sub-network to distinguish the dominant object from other distracting instances. Extensive experiments on the challenging PASCAL VOC 2012 benchmark well demonstrate the superiority of R2-IOS over other state-of-the-art methods. In particular, the APr over 20 classes at 0:5 IoU achieves 66:7%, which significantly outperforms the results of 58:7% by PFN [17] and 46:3% by [22].

This paper studies the Tensor Robust Principal Component (TRPCA) problem which extends the known Robust PCA [4] to the tensor case. Our model is based on a new tensor Singular Value Decomposition (t-SVD) [14] and its induced tensor tubal rank and tensor nuclear norm. Consider that we have a 3-way tensor X ε Rn1×n2×n3such that X = L0+ S0, where L0has low tubal rank andS0 issparse. Is that possible to recover both components? In this work, we prove that under certain suitable assumptions, we can recover both the low-rank and the sparse components exactly by simply solving a convex program whose objective is a weighted combination of the tensor nuclear norm and the l1-norm, i.e., min L, E ||L||* + λ||ε||1, s.t. X = L + ε, where λ = 1/√max(n1, n2)n3. Interestingly, TRPCA involves RPCA as a special case when n3= 1 and thus it is a simple and elegant tensor extension of RPCA. Also numerical experiments verify our theory and the application for the image denoising demonstrates the effectiveness of our method.

Modeling the aging process of human face is important for cross-age face verification and recognition. In this paper, we introduce a recurrent face aging (RFA) framework based on a recurrent neural network which can identify the ages of people from 0 to 80. Due to the lack of labeled face data of the same person captured in a long range of ages, traditional face aging models usually split the ages into discrete groups and learn a one-step face feature transformation for each pair of adjacent age groups. However, those methods neglect the in-between evolving states between the adjacent age groups and the synthesized faces often suffer from severe ghosting artifacts. Since human face aging is a smooth progression, it is more appropriate to age the face by going through smooth transition states. In this way, the ghosting artifacts can be effectively eliminated and the intermediate aged faces between two discrete age groups can also be obtained. Towards this target, we employ a twolayer gated recurrent unit as the basic recurrent module whose bottom layer encodes a young face to a latent representation and the top layer decodes the representation to a corresponding older face. The experimental results demonstrate our proposed RFA provides better aging faces over other state-of-the-art age progression methods.

We propose a nuclear-norm regularized two-dimensional neighborhood preserving projection (2DNPP) for extracting representative 2D image features. Note that 2DNPP extracts neighborhood preserving features through minimizing the reconstruction error, but the Frobenius norm based metric is sensitive to noise and outliers. To make the distance metric more reliable and model the neighborhood reconstruction error more accurately, we impose the nuclear-norm on the neighborhood reconstruction error and measure it over each image. Technically, we propose a new variant of 2DNPP termed nuclear-norm based 2DNPP (N-2DNPP). Besides, to make delivered projection promising for feature extraction, we also include the nuclear-norm constraint on projection accordingly, where the low-rank projection can embed data into their respective subspaces. Our method can outperform related state-of-the-arts in a variety of simulation settings.

In this technical demonstration, we propose a face swapping framework, which is able to interactively change the appearance of a face in the wild to a different person/creature's face in real time on a mobile device. To realize this objective, we develop a deep learning-based face detector which is able to accurately detect faces in the wild. Our face feature points tracking system based on progressive initialization ensures accurate and robust localization of facial landmarks under extreme poses and expressions in real time. Relying on the advances of our face detector and face feature points tracker, we construct the Face Swapper which can smoothly replace the face appearance of a user in real time.

Existing object proposal algorithms usually search for possible object regions over multiple locations and scales \emph{ separately}, which ignore the interdependency among different objects and deviate from the human perception procedure. To incorporate global interdependency between objects into object localization, we propose an effective Tree-structured Reinforcement Learning (Tree-RL) approach to sequentially search for objects by fully exploiting both the current observation and historical search paths. The Tree-RL approach learns multiple searching policies through maximizing the long-term reward that reflects localization accuracies over all the objects. Starting with taking the entire image as a proposal, the Tree-RL approach allows the agent to sequentially discover multiple objects via a tree-structured traversing scheme. Allowing multiple near-optimal policies, Tree-RL offers more diversity in search paths and is able to find multiple objects with a single feed-forward pass. Therefore, Tree-RL can better cover different objects with various scales which is quite appealing in the context of object proposal. Experiments on PASCAL VOC 2007 and 2012 validate the effectiveness of the Tree-RL, which can achieve comparable recalls with current object proposal algorithms via much fewer candidate windows.

In this talk, I shall introduce a series of research related with deep learning for visual understanding, and focus on three aspects: 1) how to make the model size to be small while keep high accuracy, 2) how to design proper objective functions to enhance algorithmic learnability, and 3) how to design proper network structure to accelerate the inference speed.

Research on event-based processing and analysis of media is receiving an increasing attention from the scientific community due to its relevance for an abundance of applications, from consumer video management and video surveillance to lifelogging and social media.Events have the ability to semantically encode relationships of different informational modalities, such as visual-audio-text, time, involved agents and objects, with the spatio-temporal component of events being a key feature for contextual analysis.This unveils an enormous potential for exploiting new information sources and opening new research directions.In this paper, we survey the existing literature in this field.We extensively review the employed conceptualization of the notion of event in multimedia, the techniques for event representation and modeling, the feature representation and event inference approaches for the problems of event detection in audio, visual, and textual content.Furthermore, we review some key event-based multimedia applications, and various benchmarking activities that provide solid frameworks for measuring the performance of different event processing and analysis systems.We provide an in-depth discussion of the insights obtained from reviewing the literature and identify future directions and challenges.

Human parsing, namely partitioning the human body into semantic regions, has drawn much attention recently for its wide applications in human-centric analysis. Previous works often consider solving the problem of human pose estimation as the prerequisite of human parsing. We argue that these approaches cannot obtain optimal pixel-level parsing due to the inconsistent targets between the different tasks. In this work, we directly address the problem of human parsing by using the novel Parselet representation as the building blocks of our parsing model. Parselets are a group of parsable segments which can generally be obtained by low-level over-segmentation algorithms and bear strong semantic meaning. We then build a deformable mixture parsing model (DMPM) for human parsing to simultaneously handle the deformation and multi-modalities of Parselets. The proposed model has two unique characteristics: (1) the possible numerous modalities of Parselet ensembles are exhibited as the "And-Or" structure of sub-trees; (2) to further solve the practical problem of Parselet occlusion or absence, we directly model the visibility property at some leaf nodes. The DMPM thus directly solves the problem of human parsing by searching for the best graph configuration from a pool of Parselet hypotheses without intermediate tasks. Fast rejection based on hierarchical filtering is employed to ensure the overall efficiency. Comprehensive evaluations on a new large-scale human parsing dataset, which is crawled from the Internet, with high resolution and thoroughly annotated semantic labels at pixel-level, and also a benchmark dataset demonstrate the encouraging performance of the proposed approach.

Numerous approaches on object tracking have been proposed during the past decade with demonstrated success. However, most tracking algorithms are evaluated on limited video sequences and annotations. For thorough performance evaluation, we propose a large-scale database which contains 365 challenging image sequences of pedestrians and rigid objects. The database covers 12 kinds of objects, and most of the sequences are captured from moving cameras. Each sequence is annotated with target location and occlusion level for evaluation. A thorough experimental evaluation of 20 state-of-the-art tracking algorithms is presented with detailed analysis using different metrics. The database is publicly available and evaluation can be carried out online for fair assessments of visual tracking algorithms.

The recently proposed low-rank representation (LRR) method has been empirically shown to be useful in various tasks such as motion segmentation, image segmentation, saliency detection and face recognition. While potentially powerful, LRR depends heavily on the configuration of its key parameter, λ. In realistic environments where the prior knowledge about data is lacking, however, it is still unknown how to choose λ in a suitable way. Even more, there is a lack of rigorous analysis about the success conditions of the method, and thus the significance of LRR is a little bit vague. In this paper we therefore establish a theoretical analysis for LRR, striving for figuring out under which conditions LRR can be successful, and deriving a moderately good estimate to the key parameter λ as well. Simulations on synthetic data points and experiments on real motion sequences verify our claims.

Convolutional Neural Network (CNN) has demonstrated promising performance in single-label image classification tasks. However, how CNN best copes with multi-label images still remains an open problem, mainly due to the complex underlying object layouts and insufficient multi-label training images. In this work, we propose a flexible deep CNN infrastructure, called Hypotheses-CNN-Pooling (HCP), where an arbitrary number of object segment hypotheses are taken as the inputs, then a shared CNN is connected with each hypothesis, and finally the CNN output results from different hypotheses are aggregated with max pooling to produce the ultimate multi-label predictions. Some unique characteristics of this flexible deep CNN infrastructure include: 1) no ground-truth bounding box information is required for training; 2) the whole HCP infrastructure is robust to possibly noisy and/or redundant hypotheses; 3) the shared CNN is flexible and can be well pre-trained with a large-scale single-label image dataset, e.g., ImageNet; and 4) it may naturally output multi-label prediction results. Experimental results on Pascal VOC 2007 and VOC 2012 multi-label image datasets well demonstrate the superiority of the proposed HCP infrastructure over other state-of-the-arts. In particular, the mAP reaches 90.5% by HCP only and 93.2% after the fusion with our complementary result in [44] based on hand-crafted features on the VOC 2012 dataset.

Automatically identifying characters in movies has attracted researchers' interest and led to several significant and interesting applications. However, due to the vast variation in character appearance as well as the weakness and ambiguity of available annotation, it is still a challenging problem. In this paper, we investigate this problem with the supervision of actor-character name correspondence provided by the movie cast. Our proposed framework, namely, Cast2Face, is featured by: 1) we restrict the assigned names within the set of character names in the cast; 2) for each character, by using the corresponding actor and movie name as keywords, we retrieve from the Google image search and get a group of face images to form the gallery set; 3) the probe face tracks in the movie are then identified as one of the actors by a robust kernel multitask joint sparse representation and classification method; and 4) the conditional random field model with consideration of the constraints between face tracks is introduced to enhance the final labeling. Finally, the assigned actor name of a face track is then mapped to the character name based on the cast again. Besides face naming, we further apply the proposed method to spotlight the summarization of a particular actor in his/her movies. We conduct extensive experiments and empirical evaluations on several feature-length movies to demonstrate the satisfying performance of our method.

Part-based methods have seen popular applications for face verification in the wild, since they are more robust to local variations in terms of pose, illumination, and so on. However, most of the part-based approaches are built on hand-crafted features, which may not be suitable for the specific face verification purpose. In this paper, we propose to learn a part-based feature representation under the supervision of face identities through a deep model that ensures that the generated representations are more robust and suitable for face verification. The proposed framework consists of the following two deliberate components: 1) a deep mixture model (DMM) to find accurate patch correspondence and 2) a convolutional fusion network (CFN) to extract the part-based facial features. Specifically, DMM robustly depicts the spatial-appearance distribution of patch features over the faces via several Gaussian mixtures, which provide more accurate patch correspondence even in the presence of local distortions. Then, DMM only feeds the patches which preserve the identity information to the following CFN. The proposed CFN is a two-layer cascade of convolutional neural networks: 1) a local layer built on face patches to deal with local variations and 2) a fusion layer integrating the responses from the local layer. CFN jointly learns and fuses multiple local responses to optimize the verification performance. The composite representation obtained possesses certain robustness to pose and illumination variations and shows comparable performance with the state-of-the-art methods on two benchmark data sets.

A proper strategy to alleviate overfitting is critical to a deep neural network (DNN). In this paper, we introduce the cross-loss-function regularization for boosting the generalization capability of the DNN, which results in the multi-loss regularized DNN (ML-DNN) framework. For a particular learning task, e.g., image classification, only a single-loss function is used for all previous DNNs, and the intuition behind the multiloss framework is that the extra loss functions with different theoretical motivations (e.g., pairwise loss and LambdaRank loss) may drag the algorithm away from overfitting to one particular single-loss function (e.g., softmax loss). In the training stage, we pretrain the model with the single-core-loss function and then warm start the whole ML-DNN with the convolutional parameters transferred from the pretrained model. In the testing stage, the outputs by the ML-DNN from different loss functions are fused with average pooling to produce the ultimate prediction. The experiments conducted on several benchmark datasets (CIFAR-10, CIFAR-100, MNIST, and SVHN) demonstrate that the proposed ML-DNN framework, instantiated by the recently proposed network in network, considerably outperforms all other state-of-the-art methods.

Object proposal is essential for current state-of-the-art object detection pipelines. However, the existing proposal methods generally fail in producing results with satisfying localization accuracy. The case is even worse for small objects, which, however, are quite common in practice. In this paper, we propose a novel scale-aware pixelwise object proposal network (SPOP-net) to tackle the challenges. The SPOP-net can generate proposals with high recall rate and average best overlap, even for small objects. In particular, in order to improve the localization accuracy, a fully convolutional network is employed which predicts locations of object proposals for each pixel. The produced ensemble of pixelwise object proposals enhances the chance of hitting the object significantly without incurring heavy extra computational cost. To solve the challenge of localizing objects at small scale, two localization networks, which are specialized for localizing objects with different scales are introduced, following the divide-and-conquer philosophy. Location outputs of these two networks are then adaptively combined to generate the final proposals by a large-/small-size weighting network. Extensive evaluations on PASCAL VOC 2007 and COCO 2014 show the SPOP network is superior over the state-of-the-art models. The high-quality proposals from SPOP-net also significantly improve the mean average precision of object detection with Fast-Regions with CNN features framework. Finally, the SPOP-net (trained on PASCAL VOC) shows great generalization performance when testing it on ILSVRC 2013 validation set.

Similarity-preserving hashing is a commonly used method for nearest neighbor search in large-scale image retrieval. For image retrieval, deep-network-based hashing methods are appealing, since they can simultaneously learn effective image representations and compact hash codes. This paper focuses on deep-network-based hashing for multi-label images, each of which may contain objects of multiple categories. In most existing hashing methods, each image is represented by one piece of hash code, which is referred to as semantic hashing. This setting may be suboptimal for multi-label image retrieval. To solve this problem, we propose a deep architecture that learns instance-aware image representations for multi-label image data, which are organized in multiple groups, with each group containing the features for one category. The instance-aware representations not only bring advantages to semantic hashing but also can be used in category-aware hashing, in which an image is represented by multiple pieces of hash codes and each piece of code corresponds to a category. Extensive evaluations conducted on several benchmark data sets demonstrate that for both the semantic hashing and the category-aware hashing, the proposed method shows substantial improvement over the state-of-the-art supervised and unsupervised hashing methods.

Accurate prediction of the visual attributes is significant in various recognition tasks. For many visual attributes, while it is very difficult to describe the exact degrees of their presences, by comparing the pairs of samples, the relative ordering of presences may be easily figured out. Based on this observation, instead of considering such attribute as binary attribute, the relative attribute method learns a ranking function for each attribute to provide more accurate and informative prediction results. In this paper, we also explore pairwise ranking for visual attribute prediction and propose to improve the relative attribute method in two aspects. First, we propose a relative tree method, which can achieve more accurate ranking in case of nonlinearly distributed visual data. Second, by resorting to randomization and ensemble learning, the relative tree method is extended to the relative forest method to further boost the accuracy and simultaneously reduce the computational cost. To validate the effectiveness of the proposed methods, we conduct extensive experiments on four databases: PubFig, OSR, FGNET, and WebFace. The results show that the proposed relative forest method not only outperforms the original relative attribute method, but also achieve the state-of-the-art accuracy for ordinal visual attribute prediction.

The nuclear norm is widely used as a convex surrogate of the rank function in compressive sensing for low rank matrix recovery with its applications in image recovery and signal processing. However, solving the nuclear norm-based relaxed convex problem usually leads to a suboptimal solution of the original rank minimization problem. In this paper, we propose to use a family of nonconvex surrogates of L0-norm on the singular values of a matrix to approximate the rank function. This leads to a nonconvex nonsmooth minimization problem. Then, we propose to solve the problem by an iteratively re-weighted nuclear norm (IRNN) algorithm. IRNN iteratively solves a weighted singular value thresholding problem, which has a closed form solution due to the special properties of the nonconvex surrogate functions. We also extend IRNN to solve the nonconvex problem with two or more blocks of variables. In theory, we prove that the IRNN decreases the objective function value monotonically, and any limit point is a stationary point. Extensive experiments on both synthesized data and real images demonstrate that IRNN enhances the low rank matrix recovery compared with the state-of-the-art convex algorithms.

Spectral clustering (SC) is one of the most widely used methods for data clustering. It first finds a low-dimensional embedding U of data by computing the eigenvectors of the normalized Laplacian matrix, and then performs k-means on UTto get the final clustering result. In this paper, we observe that, in the ideal case, UUTshould be block diagonal and thus sparse. Therefore, we propose the sparse SC (SSC) method that extends the SC with sparse regularization on UUT. To address the computational issue of the nonconvex SSC model, we propose a novel convex relaxation of SSC based on the convex hull of the fixed rank projection matrices. Then, the convex SSC model can be efficiently solved by the alternating direction method of multipliers Furthermore, we propose the pairwise SSC that extends SSC to boost the clustering performance by using the multi-view information of data. Experimental comparisons with several baselines on real-world datasets testify to the efficacy of our proposed methods.

In multimedia information retrieval, most classic approaches tend to represent different modalities of media in the same feature space. With the click data collected from the users' searching behavior, existing approaches take either one-to-one paired data (text-image pairs) or ranking examples (text-query-image and/or image-query-text ranking lists) as training examples, which do not make full use of the click data, particularly the implicit connections among the data objects. In this paper, we treat the click data as a large click graph, in which vertices are images/text queries and edges indicate the clicks between an image and a query. We consider learning a multimodal representation from the perspective of encoding the explicit/implicit relevance relationship between the vertices in the click graph. By minimizing both the truncated random walk loss as well as the distance between the learned representation of vertices and their corresponding deep neural network output, the proposed model which is named multimodal random walk neural network (MRW-NN) can be applied to not only learn robust representation of the existing multimodal data in the click graph, but also deal with the unseen queries and images to support cross-modal retrieval. We evaluate the latent representation learned by MRW-NN on a public large-scale click log data set Clickture and further show that MRW-NN achieves much better cross-modal retrieval performance on the unseen queries/images than the other state-of-the-art methods.

Recovering low-rank and sparse subspaces jointly for enhanced robust representation and classification is discussed. Technically, we first propose a transductive low-rank and sparse principal feature coding (LSPFC) formulation that decomposes given data into a component part that encodes low-rank sparse principal features and a noise-fitting error part. To well handle the outside data, we then present an inductive LSPFC (I-LSPFC). I-LSPFC incorporates embedded low-rank and sparse principal features by a projection into one problem for direct minimization, so that the projection can effectively map both inside and outside data into the underlying subspaces to learn more powerful and informative features for representation. To ensure that the learned features by I-LSPFC are optimal for classification, we further combine the classification error with the feature coding error to form a unified model, discriminative LSPFC (D-LSPFC), to boost performance. The model of D-LSPFC seamlessly integrates feature coding and discriminative classification, so the representation and classification powers can be enhanced. The proposed approaches are more general, and several recent existing low-rank or sparse coding algorithms can be embedded into our problems as special cases. Visual and numerical results demonstrate the effectiveness of our methods for representation and classification.

This article investigates a low-rank representation--based graph, which can used in graph-based vision tasks including image segmentation and image annotation. It naturally fuses multiple types of image features in a framework named multitask low-rank affinity pursuit. Given the image patches described with multiple types of features, we aim at inferring a unified affinity matrix that implicitly encodes the relations among these patches. This is achieved by seeking the sparsity-consistent low-rank affinities from the joint decompositions of multiple feature matrices into pairs of sparse and low-rank matrices, the latter of which is expressed as the production of the image feature matrix and its corresponding image affinity matrix. The inference process is formulated as a minimization problem and solved efficiently with the augmented Lagrange multiplier method. Considering image patches as vertices, a graph can be built based on the resulted affinity matrix. Compared to previous methods, which are usually based on a single type of feature, the proposed method seamlessly integrates multiple types of features to jointly produce the affinity matrix in a single inference step. The proposed method is applied to graph-based image segmentation and graph-based image annotation. Experiments on benchmark datasets well validate the superiority of using multiple features over single feature and also the superiority of our method over conventional methods for feature fusion.

In this article, we investigate the cross-media retrieval between images and text, that is, using image to search text (I2T) and using text to search images (T2I). Existing cross-media retrieval methods usually learn one couple of projections, by which the original features of images and text can be projected into a common latent space to measure the content similarity. However, using the same projections for the two different retrieval tasks (I2T and T2I) may lead to a tradeoff between their respective performances, rather than their best performances. Different from previous works, we propose a modality-dependent cross-media retrieval (MDCR) model, where two couples of projections are learned for different cross-media retrieval tasks instead of one couple of projections. Specifically, by jointly optimizing the correlation between images and text and the linear regression from one modal space (image or text) to the semantic space, two couples of mappings are learned to project images and text from their original feature spaces into two common latent subspaces (one for I2T and the other for T2I). Extensive experiments show the superiority of the proposed MDCR compared with other methods. In particular, based on the 4,096-dimensional convolutional neural network (CNN) visual feature and 100-dimensional Latent Dirichlet Allocation (LDA) textual feature, the mAP of the proposed method achieves the mAP score of 41.5%, which is a new state-of-the-art performance on the Wikipedia dataset.

This paper aims at developing an integrated system for clothing co-parsing (CCP), in order to jointly parse a set of clothing images (unsegmented but annotated with tags) into semantic configurations. A novel data-driven system consisting of two phases of inference is proposed. The first phase, referred as “image cosegmentation,” iterates to extract consistent regions on images and jointly refines the regions over all images by employing the exemplar-SVM technique [1]. In the second phase (i.e., “region colabeling”), we construct a multiimage graphical model by taking the segmented regions as vertices, and incorporating several contexts of clothing configuration (e.g., item locations and mutual interactions). The joint label assignment can be solved using the efficient Graph Cuts algorithm. In addition to evaluate our framework on the Fashionista dataset [2], we construct a dataset called the SYSU-Clothes dataset consisting of 2098 high-resolution street fashion photos to demonstrate the performance of our system. We achieve 90.29%/88.23% segmentation accuracy and 65.52%/63.89% recognition rate on the Fashionista and the SYSU-Clothes datasets, respectively, which are superior compared with the previous methods. Furthermore, we apply our method on a challenging task, i.e., cross-domain clothing retrieval: given user photo depicting a clothing image, retrieving the same clothing items from online shopping stores based on the fine-grained parsing results.

Along with the long-time evolution of popular social networks, e.g. Facebook, social media analysis research inevitably arrived at the era of considering face/user recognition with large age gaps. However, related research with adequate subjects and large age gaps is surprisingly rare. In this work, we first collect a so-called cross-age face (CAFE) dataset, ranging from child, to young, to adult, to old groups. Then, we propose a novel framework, called deep aging face verification (DAFV), for this challenging task. DAFV includes two modules: aging pattern synthesis and aging face verification. The aging pattern synthesis module synthesizes the faces of all age groups for the input face of an arbitrary age, and the core structure is a deep aging-aware denoising auto-encoder ( a2-DAE) with multiple outputs. The aging face verification module then takes the synthesized aging patterns of a face pair as the input, and each pair of synthesized images of the same age group is fed into a parallel CNN; finally, all parallel CNN outputs are fused to provide similar/dissimilar prediction. For DAFV, the training of the aging face verification module easily suffers from the overfitting results from the aging pattern synthesis module, and we propose to use the cross- validation strategy to produce error-aware outputs for the synthesis module. Extensive experiments on the CAFE dataset well demonstrate the superiority of the proposed DAFV framework over other solutions for aging face verification.

Feature learning is an intensively studied research topic in image classification. Although existing methods like sparse coding, locality-constrained linear coding, fisher vector encoding, etc., have shown their effectiveness in image representation, most of them overlook a phenomenon called thesmall sample size problem, where the number of training samples is relatively smaller than the dimensionality of the features, which may limit the predictive power of the classifier. Subspace learning is a strategy to mitigate this problem by reducing the dimensionality of the features. However, most conventional subspace learning methods attempt to learn a global subspace to discriminate all the classes, which proves to be difficult and ineffective in multi-class classification task. To this end, we propose to learn a local subspace for each sample instead of learning a global subspace for all samples. Our key observation is that, in multi-class image classification, the label of each testing sample is only confused by a few classes which have very similar visual appearance to it. Thus, in this work, we propose a coarse-to-fine strategy, which first picks out such classes, and then conducts a local subspace learning to discriminate them. As the subspace learning method is regularized and conducted within some selected classes, we term it selective regularized subspace learning (SRSL), and we term our classification pipeline selective regularized subspace learning based multi-class image classification (SRSL_MIC). Experimental results on four representative datasets (Caltech-101, Indoor-67, ORL Faces and AR Faces) demonstrate the effectiveness of the proposed method.

Relative attribute (RA) learning aims to learn the ranking function describing the relative strength of the attribute. Most of current learning approaches learn a linear ranking function for each attribute by use of the hand-crafted visual features. Different from the existing study, in this paper, we propose a novel deep relative attributes (DRA) algorithm to learn visual features and the effective nonlinear ranking function to describe the RA of image pairs in a unified framework. Here, visual features and the ranking function are learned jointly, and they can benefit each other. The proposed DRA model is comprised of five convolutional neural layers, five fully connected layers, and a relative loss function which contains the contrastive constraint and the similar constraint corresponding to the ordered image pairs and the unordered image pairs, respectively. To train the DRA model effectively, we make use of the transferred knowledge from the large scale visual recognition on ImageNet [1] to the RA learning task. We evaluate the proposed DRA model on three widely used datasets. Extensive experimental results demonstrate that the proposed DRA model consistently and significantly outperforms the state-of-the-art RA learning methods. On the public OSR, PubFig, and Shoes datasets, compared with the previous RA learning results [2], the average ranking accuracies have been significantly improved by about 8%, 9%, and 14%, respectively.

In this paper, we address an important and practical problem of clothing cosegmentation (CCS): given multiple fashion model photos with natural backgrounds on e-commerce websites, to automatically and simultaneously segment all images and extract the clothing regions. However, cluttered backgrounds, variations in colors and styles, and inconsistent human poses all make it a challenging task. In this paper, a novel CCS algorithm is proposed to improve the accuracy of clothing extraction by exploiting the properties of multiple clothing images with the same apparel. First, the co-salient objects are computed by detecting the upper bodies of fashion models and transferring their locations within multiple images. Based on the coarse clothing regions determined by the upper body localization and co-salient object detection, the foreground (clothing) and background Gaussian mixture models are estimated, respectively. Finally, the clothing region in each image is extracted through energy minimization based on graph cuts iteratively. The proposed cosegmentation algorithm is mainly designed for multiple clothing images. As a byproduct, it can also be applied to single image segmentation without any modification. The experiments demonstrate that the proposed approach outperforms the state-of-the-art cosegmentation methods as well as traditional single image segmentation solution for shopping images.

Saliency detection is an important procedure for machines to understand visual world as humans do. In this paper, we consider a specific saliency detection problem of predicting human eye fixations when they freely view natural images, and propose a novel dual low-rank pursuit (DLRP) method. DLRP learns saliency-aware feature transformations by utilizing available supervision information and constructs discriminative bases for effectively detecting human fixation points under the popular low-rank and sparsity-pursuit framework. Benefiting from the embedded high-level information in the supervised learning process, DLRP is able to predict fixations accurately without performing the expensive object segmentation as in the previous works. Comprehensive experiments clearly show the superiority of the proposed DLRP method over the established state-of-the-art methods. We also empirically demonstrate that DLRP provides stronger generalization performance across different data sets and inherits the advantages of both the bottom-up- and top-down-based saliency detection methods.

This work studies the Generalized Singular Value Thresholding (GSVT) operator associated with a nonconvex function g defined on the singular values of X. We prove that GSVT can be obtained by performing the proximal operator of g on the singular values since Proxg(.) is monotone when g is lower bounded. If the nonconvex g satisfies some conditions (many popular nonconvex surrogate functions, e.g., lp-norm, 0 < p < 1, of l0-norm are special cases), a general solver to find Proxg(b) is proposed for any b ≥ 0. GSVT greatly generalizes the known Singular Value Thresholding (SVT) which is a basic subroutine in many convex low rank minimization methods. We are able to solve the nonconvex low rank minimization problem by using GSVT in place of SVT.

We address the problem of describing people based on fine-grained clothing attributes. This is an important problem for many practical applications, such as identifying target suspects or finding missing people based on detailed clothing descriptions in surveillance videos or consumer photos. We approach this problem by first mining clothing images with fine-grained attribute labels from online shopping stores. A large-scale dataset is built with about one million images and fine-detailed attribute sub-categories, such as various shades of color (e.g., watermelon red, rosy red, purplish red), clothing types (e.g., down jacket, denim jacket), and patterns (e.g., thin horizontal stripes, houndstooth). As these images are taken in ideal pose/lighting/background conditions, it is unreliable to directly use them as training data for attribute prediction in the domain of unconstrained images captured, for example, by mobile phones or surveillance cameras. In order to bridge this gap, we propose a novel double-path deep domain adaptation network to model the data from the two domains jointly. Several alignment cost layers placed inbetween the two columns ensure the consistency of the two domain features and the feasibility to predict unseen attribute categories in one of the domains. Finally, to achieve a working system with automatic human body alignment, we trained an enhanced RCNN-based detector to localize human bodies in images. Our extensive experimental evaluation demonstrates the effectiveness of the proposed approach for describing people based on fine-grained clothing attributes.

Similarity-preserving hashing is a widely-used method for nearest neighbour search in large-scale image retrieval tasks. For most existing hashing methods, an image is first encoded as a vector of hand-engineering visual features, followed by another separate projection or quantization step that generates binary codes. However, such visual feature vectors may not be optimally compatible with the coding process, thus producing sub-optimal hashing codes. In this paper, we propose a deep architecture for supervised hashing, in which images are mapped into binary codes via carefully designed deep neural networks. The pipeline of the proposed deep architecture consists of three building blocks: 1) a sub-network with a stack of convolution layers to produce the effective intermediate image features; 2) a divide-and-encode module to divide the intermediate image features into multiple branches, each encoded into one hash bit; and 3) a triplet ranking loss designed to characterize that one image is more similar to the second image than to the third one. Extensive evaluations on several benchmark image datasets show that the proposed simultaneous feature learning and hash coding pipeline brings substantial improvements over other state-of-the-art supervised or unsupervised hashing methods.

This paper investigates how to perform robust and efficient unsupervised video segmentation while suppressing the effects of data noises and/or corruptions. We propose a general algorithm, called Sub-Optimal Low-rank Decomposition (SOLD), which pursues the low-rank representation for video segmentation. Given the supervoxels affinity matrix of an observed video sequence, SOLD seeks a sub-optimal solution by making the matrix rank explicitly determined. In particular, the affinity matrix with the rank fixed can be decomposed into two sub-matrices of low rank, and then we iteratively optimize them with closed-form solutions. Moreover, we incorporate a discriminative replication prior into our framework based on the obervation that small-size video patterns tend to recur frequently within the same object. The video can be segmented into several spatio-temporal regions by applying the Normalized-Cut (NCut) algorithm with the solved low-rank representation. To process the streaming videos, we apply our algorithm sequentially over a batch of frames over time, in which we also develop several temporal consistent constraints improving the robustness. Extensive experiments on the public benchmarks demonstrate superior performance of our framework over other state-of-the-art approaches.

One key challenge of facial trait recognition is the large non-rigid appearance variations due to some irrelevant real world factors, such as viewpoint and expression changes. In this paper, we explore how the shape information, i.e. facial landmark positions, can be explicitly deployed into the popular Convolutional Neural Network (CNN) architecture to disentangle such irrelevant non-rigid appearance variations. First, instead of using fixed kernels, we propose a kernel adaptation method to dynamically determine the convolutional kernels according to the spatial distribution of facial landmarks, which helps learning more robust features. Second, motivated by the intuition that different local facial regions may demand different adaptation functions, we further propose a tree-structured convolutional architecture to hierarchically fuse multiple local adaptive CNN subnetworks. Comprehensive experiments on WebFace, Morph II and MultiPIE databases well validate the effectiveness of the proposed kernel adaptation method and tree-structured convolutional architecture for facial trait recognition tasks, including identity, age and gender recognition. For all the tasks, the proposed architecture consistently achieves the state-of-the-art performances.

Both parametric and non-parametric approaches have demonstrated encouraging performances in the human parsing task, namely segmenting a human image into several semantic regions (e.g., hat, bag, left arm, face). In this work, we aim to develop a new solution with the advantages of both methodologies, namely supervision from annotated data and the flexibility to use newly annotated (possibly uncommon) images, and present a quasi-parametric human parsing model. Under the classic K Nearest Neighbor (KNN)-based nonparametric framework, the parametric Matching Convolutional Neural Network (M-CNN) is proposed to predict the matching confidence and displacements of the best matched region in the testing image for a particular semantic region in one KNN image. Given a testing image, we first retrieve its KNN images from the annotated/manually-parsed human image corpus. Then each semantic region in each KNN image is matched with confidence to the testing image using M-CNN, and the matched regions from all KNN images are further fused, followed by a superpixel smoothing procedure to obtain the ultimate human parsing result. The M-CNN differs from the classic CNN [12] in that the tailored cross image matching filters are introduced to characterize the matching between the testing image and the semantic region of a KNN image. The cross image matching filters are defined at different convolutional layers, each aiming to capture a particular range of displacements. Comprehensive evaluations over a large dataset with 7,700 annotated human images well demonstrate the significant performance gain from the quasi-parametric model over the state-of-the-arts [29, 30], for the human parsing task.

Dense local trajectories have been successfully used in action recognition. However, for most actions only a few local motion features (e.g., critical movement of hand, arm, leg etc.) are responsible for the action label. Therefore, highlighting the local features which are associated with important motion parts will lead to a more discriminative action representation. Inspired by recent advances in sentence regularization for text classification, we introduce a Motion Part Regularization framework to mine for discriminative groups of dense trajectories which form important motion parts. First, motion part candidates are generated by spatio-temporal grouping of densely extracted trajectories. Second, an objective function which encourages sparse selection for these trajectory groups is formulated together with an action class discriminative term. Then, we propose an alternative optimization algorithm to efficiently solve this objective function by introducing a set of auxiliary variables which correspond to the discriminativeness weights of each motion part (trajectory group). These learned motion part weights are further utilized to form a discriminativeness weighted Fisher vector representation for each action sample for final classification. The proposed motion part regularization framework achieves the state-of-the-art performances on several action recognition benchmarks.

Sparse representation has been applied to visual tracking by finding the best target candidate with minimal reconstruction error by use of target templates. However, most sparse representation based trackers only consider holistic or local representations and do not make full use of the intrinsic structure among and inside target candidates, thereby making the representation less effective when similar objects appear or under occlusion. In this paper, we propose a novel Structural Sparse Tracking (SST) algorithm, which not only exploits the intrinsic relationship among target candidates and their local patches to learn their sparse representations jointly, but also preserves the spatial layout structure among the local patches inside each target candidate. We show that our SST algorithm accommodates most existing sparse trackers with the respective merits. Both qualitative and quantitative evaluations on challenging benchmark image sequences demonstrate that the proposed SST algorithm performs favorably against several state-of-the-art methods.

We address the problem of cross-domain image retrieval, considering the following practical application: given a user photo depicting a clothing image, our goal is to retrieve the same or attribute-similar clothing items from online shopping stores. This is a challenging problem due to the large discrepancy between online shopping images, usually taken in ideal lighting/pose/background conditions, and user photos captured in uncontrolled conditions. To address this problem, we propose a Dual Attribute-aware Ranking Network (DARN) for retrieval feature learning. More specifically, DARN consists of two sub-networks, one for each domain, whose retrieval feature representations are driven by semantic attribute learning. We show that this attribute-guided learning is a key factor for retrieval accuracy improvement. In addition, to further align with the nature of the retrieval problem, we impose a triplet visual similarity constraint for learning to rank across the two subnetworks. Another contribution of our work is a large-scale dataset which makes the network learning feasible. We exploit customer review websites to crawl a large set of online shopping images and corresponding offline user photos with fine-grained clothing attributes, i.e., around 450,000 online shopping images and about 90,000 exact offline counterpart images of those online ones. All these images are collected from real-world consumer websites reflecting the diversity of the data modality, which makes this dataset unique and rare in the academic community. We extensively evaluate the retrieval performance of networks in different configurations. The top-20 retrieval accuracy is doubled when using the proposed DARN other than the current popular solution using pre-trained CNN features only (0.570 vs. 0.268).

Intuitive observations show that a baby may inherently possess the capability of recognizing a new visual concept (e.g., chair, dog) by learning from only very few positive instances taught by parent(s) or others, and this recognition capability can be gradually further improved by exploring and/or interacting with the real instances in the physical world. Inspired by these observations, we propose a computational model for weakly-supervised object detection, based on prior knowledge modelling, exemplar learning and learning with video contexts. The prior knowledge is modeled with a pre-trained Convolutional Neural Network (CNN). When very few instances of a new concept are given, an initial concept detector is built by exemplar learning over the deep features the pre-trained CNN. The well-designed tracking solution is then used to discover more diverse instances from the massive online weakly labeled videos. Once a positive instance is detected/identified with high score in each video, more instances possibly from different view-angles and/or different distances are tracked and accumulated. Then the concept detector can be fine-tuned based on these new instances. This process can be repeated again and again till we obtain a very mature concept detector. Extensive experiments on Pascal VOC-07/10/12 object detection datasets [9] well demonstrate the effectiveness of our framework. It can beat the state-of-the-art full-training based performances by learning from very few samples for each object category, along with about 20,000 weakly labeled videos.

In this work, we address the human parsing task with a novel Contextualized Convolutional Neural Network (Co-CNN) architecture, which well integrates the cross-layer context, global image-level context, within-super-pixel context and cross-super-pixel neighborhood context into a unified network. Given an input human image, Co-CNN produces the pixel-wise categorization in an end-to-end way. First, the cross-layer context is captured by our basic local-to-global-to-local structure, which hierarchically combines the global semantic structure and the local fine details within the cross-layers. Second, the global image-level label prediction is used as an auxiliary objective in the intermediate layer of the Co-CNN, and its outputs are further used for guiding the feature learning in subsequent convolutional layers to leverage the global image-level context. Finally, to further utilize the local super-pixel contexts, the within-super-pixel smoothing and cross-super-pixel neighbourhood voting are formulated as natural sub-components of the Co-CNN to achieve the local label consistency in both training and testing process. Comprehensive evaluations on two public datasets well demonstrate the significant superiority of our Co-CNN architecture over other state-of-the-arts for human parsing. In particular, the F-1 score on the large dataset [15] reaches 76.95% by Co-CNN, significantly higher than 62.81% and 64.38% by the state-of-the-art algorithms, M-CNN [21] and ATR [15], respectively.

In this paper, we aim to automatically render aging faces in a personalized way. Basically, a set of age-group specific dictionaries are learned, where the dictionary bases corresponding to the same index yet from different dictionaries form a particular aging process pattern cross different age groups, and a linear combination of these patterns expresses a particular personalized aging process. Moreover, two factors are taken into consideration in the dictionary learning process. First, beyond the aging dictionaries, each subject may have extra personalized facial characteristics, e.g. mole, which are invariant in the aging process. Second, it is challenging or even impossible to collect faces of all age groups for a particular subject, yet much easier and more practical to get face pairs from neighboring age groups. Thus a personality-aware coupled reconstruction loss is utilized to learn the dictionaries based on face pairs from neighboring age groups. Extensive experiments well demonstrate the advantages of our proposed solution over other state-of-the-arts in term of personalized aging progression, as well as the performance gain for cross-age face verification by synthesizing aging faces.

In this paper, we present a concise framework to approximately construct feature maps for nonlinear additive kernels such as the Intersection, Hellinger's, and X2kernels. The core idea is to construct for each individual feature a set of anchor points and assign to every query the feature map of its nearest neighbor or the weighted combination of those of its k-nearest neighbors in the anchors. The resultant feature maps can be compactly stored by a group of nearest neighbor (binary) indication vectors along with the anchor feature maps. The approximation error of such an anchored feature mapping approach is analyzed. We evaluate the performance of our approach on large-scale nonlinear support vector machines~(SVMs) learning tasks in the context of visual object classification. Experimental results on several benchmark data sets show the superiority of our method over existing feature mapping methods in achieving reasonable trade-off between training time and testing accuracy.

Feature pooling is an important strategy to achieve high performance in image classification. However, most pooling methods are unsupervised and heuristic. In this paper, we propose a novel task-driven pooling (TDP) model to directly learn the pooled representation from data in a discriminative manner. Different from the traditional methods (e.g., average and max pooling), TDP is an implicit pooling method which elegantly integrates the learning of representations into the given classification task. The optimization of TDP can equalize the similarities between the descriptors and the learned representation, and maximize the classification accuracy. TDP can be combined with the traditional BoW models (coding vectors) or the recent state-of-the-art CNN models (feature maps) to achieve a much better pooled representation. Furthermore, a self-training mechanism is used to generate the TDP representation for a new test image. A multi-task extension of TDP is also proposed to further improve the performance. Experiments on three databases (Flower-17, Indoor-67 and Caltech-101) well validate the effectiveness of our models.

Faces in the wild are usually captured with various poses, illuminations and occlusions, and thus inherently multimodally distributed in many tasks. We propose a conditional Convolutional Neural Network, named as c-CNN, to handle multimodal face recognition. Different from traditional CNN that adopts fixed convolution kernels, samples in c-CNN are processed with dynamically activated sets of kernels. In particular, convolution kernels within each layer are only sparsely activated when a sample is passed through the network. For a given sample, the activations of convolution kernels in a certain layer are conditioned on its present intermediate representation and the activation status in the lower layers. The activated kernels across layers define the sample-specific adaptive routes that reveal the distribution of underlying modalities. Consequently, the proposed framework does not rely on any prior knowledge of modalities in contrast with most existing methods. To substantiate the generic framework, we introduce a special case of c-CNN via incorporating the conditional routing of the decision tree, which is evaluated with two problems of multimodality - multi-view face identification and occluded face verification. Extensive experiments demonstrate consistent improvements over the counterparts unaware of modalities.

The beautification of human photos usually requires professional editing softwares, which are difficult for most users. In this technical demonstration, we propose a deep face beautification framework, which is able to automatically modify the geometrical structure of a face so as to boost the attractiveness. A learning based approach is adopted to capture the underlying relations between the facial shape and the attractiveness via training the Deep Beauty Predictor (DBP). Relying on the pre-trained DBP, we construct the BeAuty SHaper (BASH) to infer the "flows" of landmarks towards the maximal aesthetic level. BASH modifies the facial landmarks with the direct guidance of the beauty score estimated by DBP.

"What shall I look like after N years?" In this paper, we present an Auto Age Progression system, which automatically renders a series of aging faces in the future age ranges and generates an aging sequence (aging video) covering the entire life for an individual input. In the offline stage, a set of age-range specific dictionaries are learned from the constructed database, where the dictionary bases corresponding to the same index yet from different dictionaries form a particular aging process pattern across different age groups, and a linear combination of these patterns expresses a particular personalized aging process. In the online stage, for an input face of an individual, our system renders the aging faces corresponding to different age ranges through the aging dictionaries, and then generates an age progression by the presented face morphing technology.

We investigate how to iteratively and mutually boost object classification and detection performance by taking the outputs from one task as the context of the other one. While context models have been quite popular, previous works mainly concentrate on co-occurrence relationship within classes and few of them focus on contextualization from a top-down perspective, i.e. high-level task context. In this paper, our system adopts a new method for adaptive context modeling and iterative boosting. First, the contextualized support vector machine (Context-SVM) is proposed, where the context takes the role of dynamically adjusting the classification score based on the sample ambiguity, and thus the context-adaptive classifier is achieved. Then, an iterative training procedure is presented. In each step, Context-SVM, associated with the output context from one task (object classification or detection), is instantiated to boost the performance for the other task, whose augmented outputs are then further used to improve the former task by Context-SVM. The proposed solution is evaluated on the object classification and detection tasks of PASCAL Visual Object Classes Challenge (VOC) 2007, 2010 and SUN09 data sets, and achieves the state-of-the-art performance.

In this work, the human parsing task, namely decomposing a human image into semantic fashion/body regions, is formulated as an active template regression (ATR) problem, where the normalized mask of each fashion/body item is expressed as the linear combination of the learned mask templates, and then morphed to a more precise mask with the active shape parameters, including position, scale and visibility of each semantic region. The mask template coefficients and the active shape parameters together can generate the human parsing results, and are thus called the structure outputs for human parsing. The deep Convolutional Neural Network (CNN) is utilized to build the end-to-end relation between the input human image and the structure outputs for human parsing. More specifically, the structure outputs are predicted by two separate networks. The first CNN network is with max-pooling, and designed to predict the template coefficients for each label mask, while the second CNN network is without max-pooling to preserve sensitivity to label mask position and accurately predict the active shape parameters. For a new image, the structure outputs of the two networks are fused to generate the probability of each label for each pixel, and super-pixel smoothing is finally used to refine the human parsing result. Comprehensive evaluations on a large dataset well demonstrate the significant superiority of the ATR framework over other state-of-the-arts for human parsing. In particular, the F1-score reaches 64.38 percent by our ATR framework, significantly higher than 44.76 percent based on the state-of-the-art algorithm [28].

In this paper, we present a novel partition framework, called dense subgraph partition (DSP), to automatically, precisely and efficiently decompose a positive hypergraph into dense subgraphs. A positive hypergraph is a graph or hypergraph whose edges, except self-loops, have positive weights. We first define the concepts of core subgraph, conditional core subgraph, and disjoint partition of a conditional core subgraph, then define DSP based on them. The result of DSP is an ordered list of dense subgraphs with decreasing densities, which uncovers all underlying clusters, as well as outliers. A divide-and-conquer algorithm, called min-partition evolution, is proposed to efficiently compute the partition. DSP has many appealing properties. First, it is a nonparametric partition and it reveals all meaningful clusters in a bottom-up way. Second, it has an exact and efficient solution, called min-partition evolution algorithm. The min-partition evolution algorithm is a divide-and-conquer algorithm, thus time-efficient and memory-friendly, and suitable for parallel processing. Third, it is a unified partition framework for a broad range of graphs and hypergraphs. We also establish its relationship with the densest k-subgraph problem (DkS), an NP-hard but fundamental problem in graph theory, and prove that DSP gives precise solutions to DkS for all kin a graph-dependent set, called critical k-set. To our best knowledge, this is a strong result which has not been reported before. Moreover, as our experimental results show, for sparse graphs, especially web graphs, the size of critical k-set is close to the number of vertices in the graph. We test the proposed partition framework on various tasks, and the experimental results clearly illustrate its advantages.

In this paper, we investigate order-preserving sparse coding for classifying structured data whose atomic features possess ordering relationships. Examples include time sequences where individual frame-wise features are temporally ordered, as well as still images (landscape, street view, etc.) where different regions of the image are spatially ordered. Classification of these structured data is often tackled by first decomposing the input data into individual atomic features, then performing sparse coding or other processing for each atomic feature vector independently, and finally aggregating individual responses to classify the input data. However, this heuristic approach ignores the underlying order of the individual atomic features within the input data, and results in suboptimal discriminative capability. In this work, we introduce an order preserving regularizer which aims to preserve the ordering structure of the reconstruction coefficients within the sparse coding framework. An efficient Nesterov-type smooth approximation method is developed for optimization of the new regularization criterion, with theoretically guaranteed error bound. We perform extensive experiments for time series classification on a synthetic dataset, several machine learning benchmarks, and an RGB-D human activity dataset. We also report experiments for scene classification on a benchmark image dataset. The encoded representation is discriminative and robust, and our classifier outperforms state-of-the-art methods on these tasks.

In this paper, we present a subcategory-aware recognition framework to boost category level object classification performance. Different from the existing monolithic model approaches, we aim to automatically leverage the embedded subcategory structure to assist the further category level recognition. Motivated by the observation of considerable intra-class diversities and inter-class ambiguities in many current object classification data sets, we explicitly split data into subcategories by ambiguity-guided subcategory mining. The resulting subcategories are seamlessly integrated into the state-of-the-art detection-assisted classification framework. In particular, we build the instance affinity graph by combining both intra-class similarity and inter-class ambiguity. Visual subcategories, which correspond to the dense subgraphs, are detected by the graph shift algorithm. We then train an individual model for each subcategory rather than an attempt to represent an object category with a monolithic model. Related samples, which are informative for subcategory classification, are utilized to regularize each subcategory model. Finally, the responses from subcategory models are aggregated by subcategory-aware kernel regression. The extensive experiments over the PASCAL visual object challenge (VOC) 2007 and PASCAL VOC 2010 databases show the state-of-the-art performance from our framework.

Automated scene analysis has been a topic of great interest in computer vision and cognitive science. Recently, with the growth of crowd phenomena in the real world, crowded scene analysis has attracted much attention. However, the visual occlusions and ambiguities in crowded scenes, as well as the complex behaviors and scene semantics, make the analysis a challenging task. In the past few years, an increasing number of works on the crowded scene analysis have been reported, which covered different aspects including crowd motion pattern learning, crowd behavior and activity analyses, and anomaly detection in crowds. This paper surveys the state-of-the-art techniques on this topic. We first provide the background knowledge and the available features related to crowded scenes. Then, existing models, popular algorithms, evaluation protocols, and system performance are provided corresponding to different aspects of the crowded scene analysis. We also outline the available datasets for performance evaluation. Finally, some research problems and promising future directions are presented with discussions.

Person reidentification across nonoverlapping camera views is a rather challenging task. Due to the difficulties in obtaining identifiable faces, clothing appearance becomes the main cue for identification purposes. In this paper, we present a comprehensive study on clothing attributes assisted person reidentification. First, the body parts and their local features are extracted for alleviating the pose-misalignment issue. A latent support vector machine (LSVM)-based person reidentification approach is proposed to describe the relations among the low-level part features, middle-level clothing attributes, and high-level reidentification labels of person pairs. Motivated by the uncertainties of clothing attributes, we treat them as real-value variables instead of using them as discrete variables. Moreover, a large-scale real-world dataset with 10 camera views and about 200 subjects is collected and thoroughly annotated for this paper. The extensive experiments on this dataset show: 1) part features are more effective than features extracted from the holistic human bounding boxes; 2) the clothing attributes embedded in the LSVM model may further boost reidentification performance compared with support vector machine without clothing attributes; and 3) treating clothing attributes as real-value variables is more effective than using them as discrete variables in person reidentification.

In this paper, we present an adaptive nonparametric solution to the image parsing task, namely, annotating each image pixel with its corresponding category label. For a given test image, first, a locality-aware retrieval set is extracted from the training data based on superpixel matching similarities, which are augmented with feature extraction for better differentiation of local superpixels. Then, the category of each superpixel is initialized by the majority vote of the k -nearest-neighbor superpixels in the retrieval set. Instead of fixing k as in traditional nonparametric approaches, here, we propose a novel adaptive nonparametric approach that determines the sample-specific k for each test image. In particular, k is adaptively set to be the number of the fewest nearest superpixels that the images in the retrieval set can use to get the best category prediction. Finally, the initial superpixel labels are further refined by contextual smoothing. Extensive experiments on challenging data sets demonstrate the superiority of the new solution over other state-of-the-art nonparametric solutions.

Human action recognition is valuable for numerous practical applications, e.g., gaming, video surveillance, and video search. In this paper we hypothesize that the classification of actions can be boosted by designing a smart feature pooling strategy under the prevalently used bag-of-words-based representation. Founded on automatic video saliency analysis, we propose the spatial-temporal attention-aware pooling scheme for feature pooling. First, the video saliencies are predicted using the video saliency model, and the localized spatial-temporal features are pooled at different saliency levels and video-saliency-guided channels are formed. Saliency-aware matching kernels are thus derived as the similarity measurement of these channels. Intuitively, the proposed kernels calculate the similarities of the video foreground (salient areas) or background (nonsalient areas) at different levels. Finally, the kernels are fed into popular support vector machines for action classification. Extensive experiments on three popular data sets for action classification validate the effectiveness of our proposed method, which outperforms the state-of-the-art methods, namely 95.3% on UCF Sports (better by 4.0%), 87.9% on YouTube data set (better by 2.5%), and achieves comparable results on Hollywood2 dataset.

In this paper we address the problem of semantic segmentation. Inspired by the significant role of the context information, in this task our solution makes use of semantically meaningful overlapping object hypotheses augmented by contextual information, which is obtained from a novel background mining procedure. More precisely, a fully connected conditional random field is considered over a set of overlapping segment hypotheses and the unlabeled background regions are learned from a training set and applied in the unary terms corresponding to the foreground regions. The final segmentation result is obtained via maximum a posteriori inference, in which the segments are merged based on a sequential aggregation followed by morphological hole filling and superpixel refinement serving as postprocessing. Moreover, by incorporating other kinds of contextual cues, such as global image classification and object detection cues, a new state-of-the-art performance is achieved by our proposed solution as experimentally verified on the challenging PASCAL Visual Object Class (VOC) Challenge 2012 and MSRC-21 object segmentation data sets.

This paper addresses the problem of semantic segmentation, where the possible class labels are from a predefined set. We exploit top-down guidance, i.e., the coarse localization of the objects and their class labels provided by object detectors. For each detected bounding box, figure-ground segmentation is performed and the final result is achieved by merging the figure-ground segmentations. The main idea of the proposed approach, which is presented in our preliminary work, is to reformulate the figure-ground segmentation problem as sparse reconstruction pursuing the object mask in a nonparametric manner. The latent segmentation mask should be coherent subject to sparse error caused by intra-category diversity; thus, the object mask is inferred by making use of sparse representations over the training set. To handle local spatial deformations, local patch-level masks are also considered and inferred by sparse representations over the spatially nearby patches. The sparse reconstruction coefficients and the latent mask are alternately optimized by applying the Lasso algorithm and the accelerated proximal gradient method. The proposed formulation results in a convex optimization problem; thus, the global optimal solution is achieved. In this paper, we provide theoretical analysis of the convergence and optimality. We also give an extended numerical analysis of the proposed algorithm and a comprehensive comparison with the related semantic segmentation methods on the challenging PASCAL visual object class object segmentation datasets and the Weizmann horse dataset. The experimental results demonstrate that the proposed algorithm achieves a competitive performance when compared with the state of the arts.

To efficiently deal with the complex nonlinear variations of face images, a novel Lie group (LG) kernel is proposed in this paper to address the facial analysis problems. First, we present a linear dynamic model (LDM)-based face representation to capture both the appearance and spatial information of the face image. Second, the derived LDM can be parameterized as a specially structured upper triangular matrix, the space of which is proved to constitute an LG. An LG kernel is then designed to characterize the similarity between the LDMs for any two face images and the kernel can be fed into classical kernel-based classifiers for different types of facial analysis. Finally, experimental evaluations on face recognition and head pose estimation are conducted on several challenging data sets and the results show that the proposed algorithm outperforms other facial analysis methods.

In this paper, we study discriminative analysis of symmetric positive definite (SPD) matrices on Lie groups (LGs), namely, transforming an LG into a dimension-reduced one by optimizing data separability. In particular, we take the space of SPD matrices, e.g., covariance matrices, as a concrete example of LGs, which has proved to be a powerful tool for high-order image feature representation. The discriminative transformation of an LG is achieved by optimizing the within-class compactness as well as the between-class separability based on the popular graph embedding framework. A new kernel based on the geodesic distance between two samples in the dimension-reduced LG is then defined and fed into classical kernel-based classifiers, e.g., support vector machine, for various visual classification tasks. Extensive experiments on five public datasets, i.e., Scene-15, Caltech101, UIUC-Sport, MIT-Indoor, and VOC07, well demonstrate the effectiveness of discriminative analysis for SPD matrices on LGs, and the state-of-the-art performances are reported.

In this paper, a novel system is developed for synthesizing user-specified emotions onto arbitrary input images or videos. Other than defining the visual affective model based on empirical knowledge, a data-driven learning framework is proposed to extract the emotion-related knowledge from a set of emotion-annotated images. In a divide-and-conquer manner, the images are clustered into several emotion-specific scene subgroups for model learning. The visual affection is modeled with Gaussian mixture models based on color features of local image patches. For the purpose of affective filtering, the feature distribution of the target is aligned to the statistical model constructed from the emotion-specific scene subgroup, through a piecewise linear transformation. The transformation is derived through a learning algorithm, which is developed with the incorporation of a regularization term enforcing spatial smoothness, edge preservation, and temporal smoothness for the derived image or video transformation. Optimization of the objective function is sought via standard nonlinear method. Intensive experimental results and user studies demonstrate that the proposed affective filtering framework can yield effective and natural effects for images and videos.

This paper introduces a novel approach to facilitating image search based on a compact semantic embedding. A novel method is developed to explicitly map concepts and image contents into a unified latent semantic space for the representation of semantic concept prototypes. Then, a linear embedding matrix is learned that maps images into the semantic space, such that each image is closer to its relevant concept prototype than other prototypes. In our approach, the semantic concepts equated with query keywords and the images mapped into the vicinity of the prototype are retrieved by our scheme. In addition, a computationally efficient method is introduced to incorporate new semantic concept prototypes into the semantic space by updating the embedding matrix. This novelty improves the scalability of the method and allows it to be applied to dynamic image repositories. Therefore, the proposed approach not only narrows semantic gap but also supports an efficient image search process. We have carried out extensive experiments on various cross-modality image search tasks over three widely-used benchmark image datasets. Results demonstrate the superior effectiveness, efficiency, and scalability of our proposed approach.

We present a novel multistage learning system, called grouping estimation fusion (GEF), for human age estimation via facial images. The GEF consists of three stages: (1) age grouping; (2) age estimation within age groups; and (3) decision fusion for final age estimation. In the first stage, faces are classified into different groups, where each group has a different age range. In the second stage, three methods are adopted to extract global features from the whole face and local features from facial components (e.g., eyes, nose, and mouth). Each global or local feature is individually utilized for age estimation in each group. Thus, several decisions (i.e., estimation results) are derived. In the third stage, we obtain the final estimated age by fusing the diverse decisions from the second stage. To create diverse decisions for fusion, we investigate multiple age grouping systems in the first stage, where each system has a different number of groups and different age ranges. Thus, various decisions can be made from the second stage, and will be delivered to the third stage for fusion. Totally, six fusion schemes (i.e., intra-system fusion, inter-system fusion, intra-inter fusion, inter-intra fusion, maximum-diversity fusion, and composite fusion) are developed and compared. The performance of the GEF system is evaluated on the Face and Gesture Recognition Research Network and the MORPH-II databases, and it outperforms the existing state-of-the-art age estimation methods by a significant margin. That is, the mean absolute errors of age estimation are reduced from 4.48 to 2.81 years on FG-NET and 3.82 to 2.97 years on MORPH-II.

The challenges in visual tracking call for a method which can reliably recognize the subject of interests in an environment, where the appearance of both the background and the foreground change with time. Many existing studies model this problem as tracking by classification with online updating of the classification models, however, most of them overlook the ambiguity in visual modeling and do not consider the prior information in the tracking process. In this paper, we present a novel visual tracking method called max-confidence boosting (MCB), which explores a new way of online updating ambiguous visual phenomenon. The MCB framework models uncertainty in prior knowledge utilizing the indeterministic labels, which are used in updating models from previous frames and the new frame. Our proposed MCB tracker allows ambiguity in the tracking process and can effectively alleviate the drift problem. Many experimental results in challenging video sequences verify the success of our method, and our MCB tracker outperforms a number of the state-of-the-art tracking by classification methods.

We propose a similarity-preserving binary signature method for linear subspaces. In computer vision and pattern recognition, linear subspace is a very important representation for many kinds of data, such as face images, action and gesture videos, and so on. When there is a large amount of subspace data and the ambient dimension is high, the cost of computing the pairwise similarity between the subspaces would be high and it requires a large storage space for storing the subspaces. In this paper, we first define the angular similarity and angular distance between the subspaces. Then, based on this similarity definition, we develop a similarity-preserving binary signature method for linear subspaces, which transforms a linear subspace into a compact binary signature, and the Hamming distance between two signatures provides an unbiased estimate of the angular similarity between the two subspaces. We also provide a lower bound of the signature length sufficient to guarantee uniform distance-preservation between every pair of subspaces in a set. Experiments on face recognition, gesture recognition, and action recognition verify the effectiveness of the proposed method.

Horror content sharing on the Web is a growing phenomenon that can interfere with our daily life and affect the mental health of those involved. As an important form of expression, horror images have their own characteristics that can evoke extreme emotions. In this paper, we present a novel context-aware multi-instance learning (CMIL) algorithm for horror image recognition. The CMIL algorithm identifies horror images and picks out the regions that cause the sensation of horror in these horror images. It obtains contextual cues among adjacent regions in an image using a random walk on a contextual graph. Borrowing the strength of the fuzzy support vector machine (FSVM), we define a heuristic optimization procedure based on the FSVM to search for the optimal classifier for the CMIL. To improve the initialization of the CMIL, we propose a novel visual saliency model based on the tensor analysis. The average saliency value of each segmented region is set as its initial fuzzy membership in the CMIL. The advantage of the tensor-based visual saliency model is that it not only adaptively selects features, but also dynamically determines fusion weights for saliency value combination from different feature subspaces. The effectiveness of the proposed CMIL model is demonstrated by its use in horror image recognition on two large-scale image sets collected from the Internet.

This paper presents a general framework for solving the low-rank and/or sparse matrix minimization problems, which may involve multiple nonsmooth terms. The iteratively reweighted least squares (IRLSs) method is a fast solver, which smooths the objective function and minimizes it by alternately updating the variables and their weights. However, the traditional IRLS can only solve a sparse only or low rank only minimization problem with squared loss or an affine constraint. This paper generalizes IRLS to solve joint/mixed low-rank and sparse minimization problems, which are essential formulations for many tasks. As a concrete example, we solve the Schatten-p norm and l2,q-norm regularized low-rank representation problem by IRLS, and theoretically prove that the derived solution is a stationary point (globally optimal if p,q ≥ 1). Our convergence proof of IRLS is more general than previous one that depends on the special properties of the Schatten-p norm and l2,q-norm. Extensive experiments on both synthetic and real data sets demonstrate that our IRLS is much more efficient.

10.1145/2732265

Automatic disease inference is of importance to bridge the gap between what online health seekers with unusual symptoms need and what busy human doctors with biased expertise can offer. However, accurately and efficiently inferring diseases is non-trivial, especially for community-based health services due to the vocabulary gap, incomplete information, correlated medical concepts, and limited high quality training samples. In this paper, we first report a user study on the information needs of health seekers in terms of questions and then select those that ask for possible diseases of their manifested symptoms for further analytic. We next propose a novel deep learning scheme to infer the possible diseases given the questions of health seekers. The proposed scheme is comprised of two key components. The first globally mines the discriminant medical signatures from raw features. The second deems the raw features and their signatures as input nodes in one layer and hidden nodes in the subsequent layer, respectively. Meanwhile, it learns the inter-relations between these two layers via pre-training with pseudo-labeled data. Following that, the hidden nodes serve as raw features for the more abstract signature mining. With incremental and alternative repeating of these two components, our scheme builds a sparsely connected deep architecture with three hidden layers. Overall, it well fits specific tasks with fine-tuning. Extensive experiments on a real-world dataset labeled by online doctors show the significant performance gains of our scheme.

Human group, which indicates the people who share similar characteristics, is used to categorize humans into distinct populations or groups. In recent years, with the explosive growth of image, new concepts of human group are blooming in social networks . People in the same human group can be categorized by their facial and clothes appearance characteristics. In this work, we propose an approach to understanding the new concepts of human group with few positive samples. To this end, we construct visual models crossing two modalities related to human images and surrounding texts. Two convolutional neural networks based on face and upper body are constructed separately. Two different convolutional neural networks (CNNs) architectures are explored for visual pre-traing. To assist the human group recognition, we also merge global convolutional feature of the image. The surrounding texts are represented by semantical vectors and utilized as image labels. We transform words in the text into fixed length vectors by the skip-gram model. Then the texts corresponding to each image are converted into one feature vector by sparse coding and max pooling. Given a few positive samples of new concepts of human group, the visual model can be improved to understand the semantical meaning of the new label. The experimental results demonstrate the effectiveness of the proposed visual model and show the excellent learning capacity with few samples.

In this paper, we propose a novel semi- supervised learning strategy to address human parsing. Existing human parsing datasets are relatively small due to the required tedious human labeling. We present a general, affordable and scalable solution, which harnesses the rich contexts in those easily available web videos to boost any existing human parser. First, we crawl a large number of unlabeled videos from the web. Then for each video, the cross-frame contexts are utilized for human pose co- estimation , and then video co-parsing to obtain satisfactory human parsing results for all frames. More specifically, SIFT flow and super-pixel matching are used to build correspondences across different frames, and these correspondences then contextualize the pose estimation and human parsing in individual frames. Finally these parsed video frames are used as the reference corpus for the non-parametric human parsing component of the whole solution. To further improve the accuracy of video co-parsing, we propose an active learning method to incorporate human guidance, where the labelers are required to assess the accuracies of the pose estimation results of certain selected video frames. Then we take reliable frames as the seed frames to guide the video pose co-estimation. Our human parsing framework can then easily incorporate the human feedback to train a better fashion parser. Extensive experiments on two benchmark fashion datasets as well as a newly collected challenging Fashion Icon dataset well demonstrate the encouraging performance gain from our general pipeline for human parsing.

Feature representation learning is an important and fundamental task in multimedia and pattern recognition research. In this paper, we propose a novel framework to explore the hierarchical structure inside the images from the perspective of feature representation learning, which is applied to hierarchical image annotation. Different from the current trend in multimedia analysis of using pre-defined features or focusing on the end-task “flat” representation, we propose a novel layer-wise tag- embedded deep learning (LTDL) model to learn hierarchical features which correspond to hierarchical semantic structures in the tag hierarchy . Unlike most existing deep learning models, LTDL utilizes both the visual content of the image and the hierarchical information of associated social tags. In the training stage, the two kinds of information are fused in a bottom-up way. Supervised training and multi-modal fusion alternate in a layer-wise way to learn feature hierarchies. To validate the effectiveness of LTDL, we conduct extensive experiments for hierarchical image annotation on a large-scale public dataset. Experimental results show that the proposed LTDL can learn representative features with improved performances.

Linear subspace is an important representation for many kinds of real-world data in computer vision and pattern recognition, e.g. faces, motion videos, speeches. In this paper, first we define pairwise angular similarity and angular distance for linear subspaces. The angular distance satisfies non-negativity, identity of indiscernibles, symmetry and triangle inequality, and thus it is a metric. Then we propose a method to compress linear subspaces into compact similarity-preserving binary signatures, between which the normalized Hamming distance is an unbiased estimator of the angular distance. We provide a lower bound on the length of the binary signatures which suffices to guarantee uniform distance-preservation within a set of subspaces. Experiments on face recognition demonstrate the effectiveness of the binary signature in terms of recognition accuracy, speed and storage requirement. The results show that, compared with the exact method, the approximation with the binary signatures achieves an order of magnitude speed-up, while requiring significantly smaller amount of storage space, yet it still accurately preserves the similarity, and achieves high recognition accuracy comparable to the exact method in face recognition.

This paper proposes the Proximal Iteratively REweighted (PIRE) algorithm for solving a general problem, which involves a large body of nonconvex sparse and structured sparse related problems. Comparing with previous iterative solvers for nonconvex sparse problem, PIRE is much more general and efficient. The computational cost of PIRE in each iteration is usually as low as the state-of-the-art convex solvers. We further propose the PIRE algorithm with Parallel Splitting (PIRE-PS) and PIRE algorithm with Alternative Updating (PIRE-AU) to handle the multi-variable problems. In theory, we prove that our proposed methods converge and any limit solution is a stationary point. Extensive experiments on both synthesis and real data sets demonstrate that our methods achieve comparative learning performance, but are much more efficient, by comparing with previous nonconvex solvers.

Hashing is a popular approximate nearest neighbor search approach for large-scale image retrieval. Supervised hashing, which incorporates similarity/dissimilarity information on entity pairs to improve the quality of hashing function learning, has recently received increasing attention. However, in the existing supervised hashing methods for images, an input image is usually encoded by a vector of hand-crafted visual features. Such hand-crafted feature vectors do not necessarily preserve the accurate semantic similarities of images pairs, which may often degrade the performance of hashing function learning. In this paper, we propose a supervised hashing method for image retrieval, in which we automatically learn a good image representation tailored to hashing as well as a set of hash functions. The proposed method has two stages. In the first stage, given the pairwise similarity matrix $S$ over training images, we propose a scalable coordinate descent method to decompose $S$ into a product of $HH^T$ where $H$ is a matrix with each of its rows being the approximate hash code associated to a training image. In the second stage, we propose to simultaneously learn a good feature representation for the input images as well as a set of hash functions, via a deep convolutional network tailored to the learned hash codes in $H$ and optionally the discrete class labels of the images. Extensive empirical evaluations on three benchmark datasets with different kinds of images show that the proposed method has superior performance gains over several state-of-the-art supervised and unsupervised hashing methods.

We study the problem of human body configuration analysis, more specifically, human parsing and human pose estimation. These two tasks, ie identifying the semantic regions and body joints respectively over the human body image, are intrinsically highly correlated. However, previous works generally solve these two problems separately or iteratively. In this work, we propose a unified framework for simultaneous human parsing and pose estimation based on semantic parts. By utilizing Parselets and Mixture of Joint-Group Templates as the representations for these semantic parts, we seamlessly formulate the human parsing and pose estimation problem jointly within a unified framework via a tailored and-or graph. A novel Grid Layout Feature is then designed to effectively capture the spatial co-occurrence/occlusion information between/within the Parselets and MJGTs. Thus the mutually complementary nature of these two tasks can be harnessed to boost the performance of each other. The resultant unified model can be solved using the structure learning framework in a principled way. Comprehensive evaluations on two benchmark datasets for both tasks demonstrate the effectiveness of the proposed framework when compared with the state-of-the-art methods.

Attributes are widely used as mid-level descriptors of object properties in object recognition and retrieval. Mostly, such attributes are manually pre-defined based on domain knowledge, and their number is fixed. However, pre-defined attributes may fail to adapt to the properties of the data at hand, may not necessarily be discriminative, and/or may not generalize well. In this work, we propose a dictionary learning framework that flexibly adapts to the complexity of the given data set and reliably discovers the inherent discriminative middle-level binary features in the data. We use sample relatedness information to improve the generalization of the learned dictionary. We demonstrate that our framework is applicable to both object recognition and complex image retrieval tasks even with few training examples. Moreover, the learned dictionary also help classify novel object categories. Experimental results on the Animals with Attributes, ILSVRC2010 and PASCAL VOC2007 datasets indicate that using relatedness information leads to significant performance gains over established baselines.

The subspace segmentation problem is addressed in this paper by effectively constructing an exactly block-diagonal sample affinity matrix. The block-diagonal structure is heavily desired for accurate sample clustering but is rather difficult to obtain. Most current state-of-the-art subspace segmentation methods (such as SSC[4] and LRR[12]) resort to alternative structural priors (such as sparseness and low-rankness) to construct the affinity matrix. In this work, we directly pursue the block-diagonal structure by proposing a graph Laplacian constraint based formulation, and then develop an efficient stochastic subgradient algorithm for optimization. Moreover, two new subspace segmentation methods, the block-diagonal SSC and LRR, are devised in this work. To the best of our knowledge, this is the first research attempt to explicitly pursue such a block-diagonal structure. Extensive experiments on face clustering, motion segmentation and graph construction for semi-supervised learning clearly demonstrate the superiority of our novelly proposed subspace segmentation methods.

As surrogate functions of L0-norm, many nonconvex penalty functions have been proposed to enhance the sparse vector recovery. It is easy to extend these nonconvex penalty functions on singular values of a matrix to enhance low-rank matrix recovery. However, different from convex optimization, solving the nonconvex low-rank minimization problem is much more challenging than the nonconvex sparse minimization problem. We observe that all the existing nonconvex penalty functions are concave and monotonically increasing on [0, ∞). Thus their gradients are decreasing functions. Based on this property, we propose an Iteratively Reweighted Nuclear Norm (IRNN) algorithm to solve the nonconvex nonsmooth low-rank minimization problem. IRNN iteratively solves a Weighted Singular Value Thresholding (WSVT) problem. By setting the weight vector as the gradient of the concave penalty function, the WSVT problem has a closed form solution. In theory, we prove that IRNN decreases the objective function value monotonically, and any limit point is a stationary point. Extensive experiments on both synthetic data and real images demonstrate that IRNN enhances the low-rank matrix recovery compared with state-of-the-art convex algorithms.