Recent advances in 4D Gaussian Splatting (4DGS) editing still face challenges with view, temporal, and non-editing region consistency, as well as with handling complex text instructions. To address these issues, we propose4DGS-Craft, a consistent and interactive 4DGS editing framework. We first introduce a 4D-aware InstructPix2Pix model to ensure both view and temporal consistency. This model incorporates 4D VGGT geometry features extracted from the initial scene, enabling it to capture underlying 4D geometric structures during editing. We further enhance this model with a multi-view grid module that enforces consistency by iteratively refining multi-view input images while jointly optimizing the underlying 4D scene. Furthermore, we preserve the consistency of non-edited regions through a novel Gaussian selection mechanism, which identifies and optimizes only the Gaussians within the edited regions. Beyond consistency, facilitating user interaction is also crucial for effective 4DGS editing. Therefore, we design an LLM-based module for user intent understanding. This module employs a user instruction template to define atomic editing operations and leverages an LLM for reasoning. As a result, our framework can interpret user intent and decompose complex instructions into a logical sequence of atomic operations, enabling it to handle intricate user commands and further enhance editing performance. Compared to related works, our approach enables more consistent and controllable 4D scene editing.

Robotic manipulation requires learning a generalizable policy that can adapt to complicated new environments. However, existing methods typically overlook the inherent task complexity and employ a policy with the same budget for tasks with varied difficulties, facing challenges in inefficient computational resource allocation and zero-shot generalization. In this work, we identify three facets of complexity imbalance issues in the current manipulation tasks at the Inter-task, Intra-task, and Noise-timesteps levels. To address this gap, we introduce the Complexity-Aware Policy (CAP), a novel approach integrating flow matching with a Transformer-based backbone and a Mixture of Heterogeneous Experts (MoHE) structure for policy learning. By leveraging Rectified Flow and dynamically adjusting model capacity based on task complexity, which is assessed through features like object counts and precision needs, our method allocates computational resources efficiently and effectively. This results in faster convergence, optimized computational resource usage, and improved precision across diverse manipulation tasks. Our proposed method achieves the state-of-the-art performance on widely-used CALVIN, LIBERO, and SimplerEnv benchmarks, and is further validated through six real-world experiments, where it consistently outperforms baseline methods across all tasks.

Diffusion-based policies demonstrate remarkable capabilities in generating robust and precise actions for embodied tasks. However, when adapting to a new environment with visual domain gap such as lighting variation, object appearance change, different background, and unseen distractors, these methods require a substantial amount of human demonstrations gathered through teleoperation. This data collection process incurs significant costs in terms of both time and financial resources. To this end, we introduceFreeHand, which enables cross-environment adaptation without requiring additional teleoperated demonstrations,freeing human handsfrom the time-consuming data collection process. Specifically, we find that a well-trained diffusion policy network struggles to adapt to new environments due to distribution mismatches at both the visual and policy levels. Simply aligning domains at the visual level is insufficient, as even subtle visual changes in the environment can lead to severe action failures. Therefore, we propose a two-level alignment scheme. The visual-level alignment is achieved through adversarial training between the visual features from old and new environments. For the policy-level alignment, we introduce a novel noise-aware adversarial learning strategy for the diffusion policy. We validate our approach through comprehensive cross-domain adaptation experiments under three settings: Sim2Sim (PushT, CALVIN, RoboTwin), Real2Sim (SimplerEnv), and Real2Real (real-world robotic deployment). Our method demonstrates significant improvement with minimal additional parameter overhead, showcasing its effectiveness and efficiency.

Diffusion models excel at image generation. Recent studies have shown that these models not only generate high-quality images but also encode text-image alignment information through attention maps or loss functions. This information is valuable for various downstream tasks, including segmentation, text-guided image editing, and compositional image generation. However, current methods heavily rely on the assumption of perfect text-image alignment in diffusion models, which is not the case. In this paper, we propose using zero-shot referring image segmentation as a proxy task to evaluate the pixel-level image and class-level text alignment of popular diffusion models. We conduct an in-depth analysis of pixel-text misalignment in diffusion models from the perspective of training data bias. We find that misalignment occurs in images with small-sized, occluded, or rare object classes. Therefore, we propose ELBO-T2IAlign-a simple yet effective method to calibrate pixel-text alignment in diffusion models based on the evidence lower bound (ELBO) of likelihood. ELBO-T2IAlign is training-free and generic: it requires no additional annotations, model retraining, or architectural modifications, and it can be directly applied to different diffusion backbones. Extensive experiments on zero-shot referring image segmentation, text-guided image editing, and compositional image generation verify that the proposed calibration improves pixel-text alignment across complementary downstream tasks.

The unprecedented advancements in Large Language Models (LLMs) have profoundly impacted natural language processing but have yet to fully embrace the realm of scalable vector graphics (SVG) generation. While LLMs encode partial knowledge of SVG data from web pages during training, recent findings suggest that semantically ambiguous and tokenized representations within LLMs may result in hallucinations in vector primitive predictions. Additionally, LLM training typically lacks modeling and understanding of the rendering sequence of vector paths, which can lead to occlusion between output vector primitives. In this paper, we present LLM4SVG, an initial yet substantial step toward bridging this gap by enabling LLMs to better understand and generate vector graphics. LLM4SVG facilitates a deeper understanding of SVG components through learnable semantic tokens, which precisely encode these tokens and their corresponding properties to generate semantically aligned SVG outputs. Using a series of learnable semantic tokens, a structured dataset for instruction following is developed to support comprehension and generation across two primary tasks. Our method introduces a modular architecture to existing large language models, integrating semantic tags, vector instruction encoders, fine-tuned commands, and powerful LLMs to tightly combine geometric, appearance, and language information. To overcome the scarcity of SVG-text instruction data, we developed an automated data generation pipeline that collected our SVGX-SFT Dataset, consisting of high-quality human-designed SVGs and 580k SVG instruction following data specifically crafted for LLM training, which facilitated the adoption of the supervised fine-tuning strategy popular in LLM development. By exploring various training strategies, we developed LLM4SVG, which significantly moves beyond optimized rendering-based approaches and language-model-based baselines to achieve remarkable results in human evaluation tasks. Code, model, and data will be released at: https://ximinng.github.io/LLM4SVGProject/

Diffusion Policies have significantly advanced robotic manipulation tasks via imitation learning, but their application on resource-constrained mobile platforms remains challenging due to computational inefficiency and extensive memory footprint. In this paper, we propose LightDP, a novel framework specifically designed to accelerate Diffusion Policies for real-time deployment on mobile devices. LightDP addresses the computational bottleneck through two core strategies: network compression of the denoising modules and reduction of the required sampling steps. We first conduct an extensive computational analysis on existing Diffusion Policy architectures, identifying the denoising network as the primary contributor to latency. To overcome performance degradation typically associated with conventional pruning methods, we introduce a unified pruning and retraining pipeline, optimizing the model's post-pruning recoverability explicitly. Furthermore, we combine pruning techniques with consistency distillation to effectively reduce sampling steps while maintaining action prediction accuracy. Experimental evaluations on the standard datasets, \ie, PushT, Robomimic, CALVIN, and LIBERO, demonstrate that LightDP achieves real-time action prediction on mobile devices with competitive performance, marking an important step toward practical deployment of diffusion-based policies in resource-limited environments. Extensive real-world experiments also show the proposed LightDP can achieve performance comparable to state-of-the-art Diffusion Policies.

The high computational cost and slow inference time are major obstacles to deploying Video Diffusion Models (VDMs). To overcome this, we introduce a new Video Diffusion Model Compression approach using individual content and motion dynamics preserved pruning and consistency loss. First, we empirically observe that deeper VDM layers are crucial for maintaining the quality of motion dynamics (e.g., coherence of the entire video), while shallower layers are more focused on individual content (e.g., individual frames). Therefore, we prune redundant blocks from the shallower layers while preserving more of the deeper layers, resulting in a lightweight VDM variant called VDMini. Moreover, we propose an Individual Content and Motion Dynamics (ICMD) Consistency Loss to gain comparable generation performance as larger VDM to VDMini. In particular, we first use the Individual Content Distillation (ICD) Loss to preserve the consistency in the features of each generated frame between the teacher and student models. Next, we introduce a Multi-frame Content Adversarial (MCA) Loss to enhance the motion dynamics across the generated video as a whole. This method significantly accelerates inference time while maintaining high-quality video generation. Extensive experiments demonstrate the effectiveness of our VDMini on two important video generation tasks, Text-to-Video (T2V) and Image-to-Video (I2V), where we respectively achieve an average 2.5 ×, 1.4 ×, and 1.25 × speed up for the I2V method SF-V, the T2V method T2V-Turbo-v2, and the T2V method HunyuanVideo, while maintaining the quality of the generated videos on several benchmarks including UCF101, VBench-T2V, and VBench-I2V.

The goal of this workshop is to showcase the latest advancements in generative AI (GAI) for creating, editing, restoring, and compressing rich media data, including images, videos, and 3D content. GAI models such as VAEs, GANs, and diffusion models have demonstrated remarkable impact in both academic research and industrial applications. For example, GAI enables users to design and generate synthetic yet realistic content without requiring professional artistic or technical expertise, driving significant market growth in gaming and entertainment. Beyond creative applications, GAI also provides crucial simulated data for training embodied AI agents. When applied to media restoration and synthesis, GAI techniques can further alleviate transmission challenges by offloading computation to client devices. To advance this field, the workshop will host four competition tracks using novel industry-level data, solicit high-quality paper submissions, and invite leading speakers from academia and industry to foster collaboration and innovation. In particular, the competition focuses on media generation and transmission with GAI. The first three tracks address reducing computation and transmission costs for efficient media delivery, while the fourth track focuses on controlled novel content creation. To support these challenges, a large-scale multi-modality, multi-view dataset named M3VIR is provided. This dataset comprises a diverse collection of videos simulated using the UE5 Unreal Engine, with carefully matched content serving as ground truth for the competition tasks.

3D Gaussian Splatting (3DGS) data compression is crucial for enabling efficient storage and transmission in 3D scene modeling. However, its development remains limited due to inadequate entropy models and suboptimal quantization strategies for both lossless and lossy compression scenarios, where existing methods have yet to 1) fully leverage hyperprior information to construct robust conditional entropy models, and 2) apply fine-grained, element-wise quantization strategies for improved compression granularity. In this work, we propose a novel Mixture of Priors (MoP) strategy to simultaneously address these two challenges. Specifically, inspired by the Mixture-of-Experts (MoE) paradigm, our MoP approach processes hyperprior information through multiple lightweight MLPs to generate diverse prior features, which are subsequently integrated into the MoP feature via a gating mechanism. To enhance lossless compression, the resulting MoP feature is utilized as a hyperprior to improve conditional entropy modeling. Meanwhile, for lossy compression, we employ the MoP feature as guidance information in an element-wise quantization procedure, leveraging a prior-guided Coarse-to-Fine Quantization (C2FQ) strategy with a predefined quantization step value. Specifically, we expand the quantization step value into a matrix and adaptively refine it from coarse to fine granularity, guided by the MoP feature, thereby obtaining a quantization step matrix that facilitates element-wise quantization. Extensive experiments demonstrate that our proposed 3DGS data compression framework achieves state-of-the-art performance across multiple benchmarks, including Mip-NeRF360, BungeeNeRF, DeepBlending, and Tank&Temples.

In this work, we introduce CAD-Coder, a novel framework that reformulates text-to-CAD as the generation of CadQuery scripts—a Python-based, parametric CAD language. This representation enables direct geometric validation, a richer modeling vocabulary, and seamless integration with existing LLMs. To further enhance code validity and geometric fidelity, we propose a two-stage learning pipeline: (1) supervised fine-tuning on paired text–CadQuery data, and (2) reinforcement learning with Group Reward Policy Optimization (GRPO), guided by a CAD-specific reward comprising both a geometric reward (Chamfer Distance) and a format reward. We also introduce a chain-of-thought (CoT) planning process to improve model reasoning, and construct a large-scale, high-quality dataset of 110K text–CadQuery–3D model triplets and 1.5K CoT samples via an automated pipeline. Extensive experiments demonstrate that CAD-Coder enables LLMs to generate diverse, valid, and complex CAD models directly from natural language, advancing the state of the art of text-to-CAD generation and geometric reasoning.

Recently, text-guided scalable vector graphics (SVG) synthesis has shown great promise in domains like iconography and sketching. However, existing Text-to-SVG methods often face challenges in editability, visual quality, and diversity. To address these issues, we propose a novel framework for text-guided SVG synthesis that significantly enhances editability, quality, and diversity. To enhance the editability of output SVGs, we introduce a Hierarchical Image VEctorization (HIVE) framework that operates at the semantic object level and supervises the optimization of components within the vector object. This approach facilitates the decoupling of vector graphics into distinct objects and component levels. Our proposed HIVE algorithm, informed by image segmentation priors, not only ensures a more precise representation of vector graphics but also enables fine-grained editing capabilities within vector objects. To improve the diversity of output SVGs, we present a Vectorized Particle-based Score Distillation (VPSD) approach. VPSD addresses over-saturation issues in existing methods and enhances sample diversity. A pre-trained reward model is incorporated to re-weight vector particles, improving aesthetic appeal and enabling faster convergence. Additionally, we design a novel adaptive vector primitives control strategy, which allows for the dynamic adjustment of the number of primitives, thereby enhancing the presentation of graphic details. Extensive experiments validate the effectiveness of the proposed method, demonstrating its superiority over baseline methods in terms of editability, visual quality, and diversity. We also show that our new method supports up to six distinct vector styles, capable of generating high-quality vector assets suitable for stylized vector design and poster design.

The pre-trained text-image discriminative models, such as CLIP, has been explored for open-vocabulary semantic segmentation with unsatisfactory results due to the loss of crucial localization information and awareness of object shapes. Recently, there has been a growing interest in expanding the application of generative models from generation tasks to semantic segmentation. These approaches utilize generative models either for generating annotated data or extracting features to facilitate semantic segmentation. This typically involves generating a considerable amount of synthetic data or requiring additional mask annotations. To this end, we uncover the potential of generative text-to-image diffusion models (e.g., Stable Diffusion) as highly efficient open-vocabulary semantic segmenters, and introduce a novel training-free approach named DiffSegmenter. The insight is that to generate realistic objects that are semantically faithful to the input text, both the complete object shapes and the corresponding semantics are implicitly learned by diffusion models. We discover that the object shapes are characterized by the self-attention maps while the semantics are indicated through the cross-attention maps produced by the denoising U-Net, forming the basis of our segmentation results. Additionally, we carefully design effective textual prompts and a category filtering mechanism to further enhance the segmentation results. Extensive experiments on three benchmark datasets show that the proposed DiffSegmenter achieves impressive results for open-vocabulary semantic segmentation.

3D point cloud semantic segmentation has a wide range of applications. Recently, weakly supervised point cloud segmentation methods have been proposed, aiming to alleviate the expensive and laborious manual annotation process by leveraging scene-level labels. However, these methods have not effectively exploited the rich geometric information (such as shape and scale) and appearance information (such as color and texture) present in RGB-D scans. Furthermore, current approaches fail to fully leverage the point affinity that can be inferred from the feature extraction network, which is crucial for learning from weak scene-level labels. Additionally, previous work overlooks the detrimental effects of the long-tailed distribution of point cloud data in weakly supervised 3D semantic segmentation. To this end, this paper proposes a simple yet effective scene-level weakly supervised point cloud segmentation method with a newly introduced multi-modality point affinity inference module. The point affinity proposed in this paper is characterized by features from multiple modalities (e.g., point cloud and RGB), and is further refined by normalizing the classifier weights to alleviate the detrimental effects of long-tailed distribution without the need of the prior of category distribution. Extensive experiments on the ScanNet and S3DIS benchmarks verify the effectiveness of our proposed method, which outperforms the state-of-the-art by ~4% to ~ 6% mIoU. Codes are released at https://github.com/Sunny599/AAAI24-3DWSSG-MMA.

Conventional Federated Domain Adaptation (FDA) approaches usually demand an abundance of assumptions, which makes them significantly less feasible for real-world situations and introduces security hazards. This paper relaxes the assumptions from previous FDAs and studies a more practical scenario named Universal Federated Domain Adaptation (UFDA). It only requires the black-box model and the label set information of each source domain, while the label sets of different source domains could be inconsistent, and the target-domain label set is totally blind. Towards a more effective solution for our newly proposed UFDA scenario, we propose a corresponding methodology called Hot-Learning with Contrastive Label Disambiguation (HCLD). It particularly tackles UFDA's domain shifts and category gaps problems by using one-hot outputs from the black-box models of various source domains. Moreover, to better distinguish the shared and unknown classes, we further present a cluster-level strategy named Mutual-Voting Decision (MVD) to extract robust consensus knowledge across peer classes from both source and target domains. Extensive experiments on three benchmark datasets demonstrate that our method achieves comparable performance for our UFDA scenario with much fewer assumptions, compared to previous methodologies with comprehensive additional assumptions.

Deep learning models have the ability to extract rich knowledge from large-scale datasets. However, the sharing of data has become increasingly challenging due to concerns regarding data copyright and privacy. Consequently, this hampers the effective transfer of knowledge from existing data to novel downstream tasks and concepts. Zero-shot learning (ZSL) approaches aim to recognize new classes by transferring semantic knowledge learned from base classes. However, traditional generative ZSL methods often require access to real images from base classes and rely on manually annotated attributes, which presents challenges in terms of data restrictions and model scalability. To this end, this paper tackles a challenging and practical problem dubbed as data-free zero-shot learning (DFZSL), where only the CLIP-based base classes data pre-trained classifier is available for zero-shot classification. Specifically, we propose a generic framework for DFZSL, which consists of three main components. Firstly, to recover the virtual features of the base data, we model the CLIP features of base class images as samples from a von Mises-Fisher (vMF) distribution based on the pre-trained classifier. Secondly, we leverage the text features of CLIP as low-cost semantic information and propose a feature-language prompt tuning (FLPT) method to further align the virtual image features and textual features. Thirdly, we train a conditional generative model using the well-aligned virtual image features and corresponding semantic text features, enabling the generation of new classes features and achieve better zero-shot generalization. Our framework has been evaluated on five commonly used benchmarks for generalized ZSL, as well as 11 benchmarks for the base-to-new ZSL. The results demonstrate the superiority and effectiveness of our approach. Our code is available in https://github.com/ylong4/DFZSL.

Recently, text-guided scalable vector graphics (SVGs) synthesis has shown promise in domains such as iconography and sketch. However, existing text-to-SVG generation methods lack editability and struggle with visual quality and result diversity. To address these limitations, we propose a novel text-guided vector graphics synthesis method called SVGDreamer. SVGDreamer incorporates a semantic-driven image vectorization (SIVE) process that enables the decomposition of synthesis into foreground objects and background, thereby enhancing editability. Specifically, the SIVE process introduces attention-based primitive control and an attention-mask loss function for effective control and manipulation of individual elements. Additionally, we propose a Vectorized Particle-based Score Distillation (VPSD) approach to address issues of shape over-smoothing, color over-saturation, limited diversity, and slow convergence of the existing text-to-SVG generation methods by modeling SVGs as distributions of control points and colors. Furthermore, VPSD leverages a reward model to re-weight vector particles, which improves aesthetic appeal and accelerates convergence. Extensive experiments are conducted to validate the effectiveness of SVGDreamer, demonstrating its superiority over baseline methods in terms of editability, visual quality, and diversity. Project page: https://ximinng.github.io/SVGDreamer-project/

Content-adaptive compression is crucial for enhancing the adaptability of the pre-trained neural codec for various contents. Though, its application in neural video compression (NVC) is still limited due to two main aspects: 1), video compression relies heavily on temporal redundancy, therefore updating just one or a few frames can lead to significant errors accumulating over time; 2), NVC frameworks are generally complex, with many comprehensive components that are not trivial to update quickly during the encoding procedure. To address these challenges, we have developed a content-adaptive NVC technique called Group-aware Parameter-efficient Updating (GPU). Initially, to minimize error accumulation, we adopt a group-aware approach for updating encoder parameters. This involves adopting a patch-based Group of Pictures (GoP) updating strategy to segment a video into patch-based GoPs, which will be updated to facilitate a globally optimized domain-transferable solution. Subsequently, we introduce a parameter-efficient delta-tuning strategy, which is achieved by integrating several light-weight adapters into each encoding component by using both serial and parallel configuration. Such architecture-agnostic modules stimulate the components with large parameters, thereby reducing the updating cost during the encoding stage. We incorporate our GPU into the latest NVC framework and conduct extensive experiments, whose results showcase outstanding video compression efficiency across six video compression benchmarks and the adaptability of one medical volumetric image compression benchmark.

Reconstructing a 3D shape based on a single sketch image is challenging due to the inherent sparsity and ambiguity present in sketches. Existing methods lose fine details when extracting features to predict 3D objects from sketches. Upon analyzing the 3D-to-2D projection process, we observe that the density map, characterizing the distribution of 2D point clouds, can serve as a proxy to facilitate the reconstruction process. In this work, we propose a novel sketch-based 3D reconstruction model named SketchSampler. It initiates the process by translating a sketch through an image translation network into a more informative 2D representation, which is then used to generate a density map. Subsequently, a two-stage probabilistic sampling process is employed to reconstruct a 3D point cloud: first, recovering the 2D points (i.e., the x and y coordinates) by sampling the density map; and second, predicting the depth (i.e., the z coordinate) by sampling the depth values along the ray determined by each 2D point. Additionally, we convert the reconstructed point cloud into a 3D mesh for wider applications. To reduce ambiguity, we incorporate hidden lines in sketches. Experimental results demonstrate that our proposed approach significantly outperforms other baseline methods.

Object parts serve as crucial intermediate representations in various downstream tasks, but part-level representation learning still has not received as much attention as other vision tasks. Previous research has established that Vision Transformer can learn instance-level attention without labels, extracting high-quality instance-level representations for boosting downstream tasks. In this paper, we achieve unsupervised part-specific attention learning using a novel paradigm and further employ the part representations to improve part discovery performance. Specifically, paired images are generated from the same image with different geometric transformations, and multiple part representations are extracted from these paired images using a novel module, named PartFormer. These part representations from the paired images are then exchanged to improve geometric transformation invariance. Subsequently, the part representations are aligned with the feature map extracted by a feature map encoder, achieving high similarity with the pixel representations of the corresponding part regions and low similarity in irrelevant regions. Finally, the geometric and semantic constraints are applied to the part representations through the intermediate results in alignment for part-specific attention learning, encouraging the PartFormer to focus locally and the part representations to explicitly include the information of the corresponding parts. Moreover, the aligned part representations can further serve as a series of reliable detectors in the testing phase, predicting pixel masks for part discovery. Extensive experiments are carried out on four widely used datasets, and our results demonstrate that the proposed method achieves competitive performance and robustness due to its part-specific attention.

Detection transformers such as DETR [1] have recently exhibited promising performance for many object detection tasks, but the generalization ability of those methods is still quite limited for cross-domain adaptation scenarios. To address the cross-domain issue, a straightforward method is to perform token alignment with adversarial training in transformers. However, its performance is often unsatisfactory because the tokens in detection transformers are quite diverse and represent different spatial and semantic information. In this paper, we propose a new method for cross-domain detection transformers called spatial-aware and semantic-aware token alignment (SSTA). Specifically, we take advantage of the characteristics of cross-attention as used in the detection transformer and propose spatial-aware token alignment (SpaTA) and semantic-aware token alignment (SemTA) strategies to guide the token alignment across domains. For spatial-aware token alignment, we extract the information from the cross-attention map (CAM) to align the distribution of tokens according to their attention to object queries. For semantic-aware token alignment, we inject the category information into the cross-attention map and construct domain embedding to guide the learning of a multi-class discriminator to model the category relationship and achieve category-level token alignment during the entire adaptation process. We conduct extensive experiments on several widely-used benchmarks, and the results clearly show the effectiveness of our proposed approach over existing state-of-the-art methods.

Observing that the existing model compression approaches only focus on reducing the redundancies in convolutional neural networks (CNNs) along one particular dimension (e.g., the channel or spatial or temporal dimension), in this work, we propose our multidimensional pruning (MDP) framework, which can compress both 2-D CNNs and 3-D CNNs along multiple dimensions in an end-to-end fashion. Specifically, MDP indicates the simultaneous reduction of channels and more redundancy on other additional dimensions. The redundancy of additional dimensions depends on the input data, i.e., spatial dimension for 2-D CNNs when using images as the input data, and spatial and temporal dimensions for 3-D CNNs when using videos as the input data. We further extend our MDP framework to the MDP-Point approach for compressing point cloud neural networks (PCNNs) whose inputs are irregular point clouds (e.g., PointNet). In this case, the redundancy along the additional dimension indicates the point dimension (i.e., the number of points). Comprehensive experiments on six benchmark datasets demonstrate the effectiveness of our MDP framework and its extended version MDP-Point for compressing CNNs and PCNNs, respectively.

In this work, we propose the complexity-guided slimmable decoder (cgSlimDecoder) in combination with skip-adaptive entropy coding (SaEC) for efficient deep video compression. Specifically, given the target complexty constraints, in our cgSlimDecoder, we introduce a set of new channel width selection modules to automatically decide the optimal channel width of each slimmable convolution layer. By optimizing the complexity-rate-distortion related objective function to directly learn the parameters of the newly introduced channel width selection modules and other modules in the decoder, our cgSlimDecoder can automatically allocate the optimal numbers of parameters for different types of modules (e.g., motion/residual decoder and the motion compensation network) and simultaneously support multiple complexity levels by using a single learnt decoder instead of multiple decoders. In addition, our proposed SaEC can further accelerate the entropy decoding procedure in both motion and residual decoders by simply skipping the entropy coding process for the elements in the encoded feature maps that are already well-predicted by the hyperprior network. As demonstrated in our comprehensive experiments, our newly proposed methods cgSlimDecoder and SaEC are general and can be readily incorporated into three widely used deep video codecs (i.e., DVC, FVC and DCVC) to significantly improve their coding efficiency with negligible performance drop.

The task of 3D semantic scene graph (3D SSG) prediction in the point cloud is challenging since (1) the 3D point cloud only captures geometric structures with limited semantics compared to 2D images, and (2) long-tailed relation distribution inherently hinders the learning of unbiased prediction. Since 2D images provide rich semantics and scene graphs are in nature coped with languages, in this study, we propose Visual-Linguistic Semantics Assisted Training (VL-SAT) scheme that can significantly empower 3DSSG prediction models with discrimination about long-tailed and ambiguous semantic relations. The key idea is to train a powerful multi-modal oracle model to assist the 3D model. This oracle learns reliable structural representations based on semantics from vision, language, and 3D geometry, and its benefits can be heterogeneously passed to the 3D model during the training stage. By effectively utilizing visual-linguistic semantics in training, our VL-SAT can significantly boost common 3DSSG prediction models, such as SGFN and SGGpoint, only with 3D inputs in the inference stage, especially when dealing with tail relation triplets. Comprehensive evaluations and ablation studies on the 3DSSG dataset have validated the effectiveness of the proposed scheme. Code is available at https://github.com/wz7in/CVPR2023-VLSAT.

Semi-supervised semantic segmentation (SSS) has recently gained increasing research interest as it can reduce the requirement for large-scale fully-annotated training data. The current methods often suffer from the confirmation bias from the pseudo-labelling process, which can be alleviated by the co-training framework. The current co-training-based SSS methods rely on hand-crafted perturbations to prevent the different sub-nets from collapsing into each other, but these artificial perturbations cannot lead to the optimal solution. In this work, we propose a new conflict-based cross-view consistency (CCVC) method based on a two-branch co-training framework which aims at enforcing the two sub-nets to learn informative features from irrelevant views. In particular, we first propose a new cross-view consistency (CVC) strategy that encourages the two sub-nets to learn distinct features from the same input by introducing a feature discrepancy loss, while these distinct features are expected to generate consistent prediction scores of the input. The CVC strategy helps to prevent the two sub-nets from stepping into the collapse. In addition, we further propose a conflict-based pseudo-labelling (CPL) method to guarantee the model will learn more useful information from conflicting predictions, which will lead to a stable training process. We validate our new CCVC approach on the SSS benchmark datasets where our method achieves new state-of-the-art performance. Our code is available at https://github.com/xiaoyao3302/CCVC.

Although existing neural video compression~(NVC) methods have achieved significant success, most of them focus on improving either temporal or spatial information separately. They generally use simple operations such as concatenation or subtraction to utilize this information, while such operations only partially exploit spatio-temporal redundancies. This work aims to effectively and jointly leverage robust temporal and spatial information by proposing a new 3D-based transformer module: Spatio-Temporal Cross-Covariance Transformer (ST-XCT). The ST-XCT module combines two individual extracted features into a joint spatio-temporal feature, followed by 3D convolutional operations and a novel spatio-temporal-aware cross-covariance attention mechanism. Unlike conventional transformers, the cross-covariance attention mechanism is applied across the feature channels without breaking down the spatio-temporal features into local tokens. Such design allows for modeling global cross-channel correlations of the spatio-temporal context while lowering the computational requirement. Based on ST-XCT, we introduce a novel transformer-based end-to-end optimized NVC framework. ST-XCT-based modules are integrated into various key coding components of NVC, such as feature extraction, frame reconstruction, and entropy modeling, demonstrating its generalizability. Extensive experiments show that our ST-XCT-based NVC proposal achieves state-of-the-art compression performances on various standard video benchmark datasets.

Neural image compression has gained significant attention thanks to the remarkable success of deep neural networks. However, most existing neural image codecs focus solely on improving human vision perception. In this work, our objective is to enhance image compression methods for both human vision quality and machine vision tasks simultaneously. To achieve this, we introduce a novel approach to Partition, Transmit, Reconstruct, and Aggregate (PTRA) the latent representation of images to balance the optimizations for both aspects. By employing our method as a module in existing neural image codecs, we create a latent representation predictor that dynamically manages the bit-rate cost for machine vision tasks. To further improve the performance of auto-regressive-based coding techniques, we enhance our hyperprior network and predictor module with context modules, resulting in a reduction in bit-rate. The extensive experiments conducted on various machine vision benchmarks such as ILSVRC 2012, VOC 2007, VOC 2012, and COCO demonstrate the superiority of our newly proposed image compression framework. It outperforms existing neural image compression methods in multiple machine vision tasks including classification, segmentation, and detection, while maintaining high-quality image reconstruction for human vision.

With the emergence of VR and AR, 360° data attracts increasing attention from the computer vision and multimedia communities. Typically, 360° data is projected into 2D ERP (equirectangular projection) images for feature extraction. However, existing methods cannot handle the distortions that result from the projection, hindering the development of 360-data-based tasks. Therefore, in this paper, we propose a Transformer-based model called DATFormer to address the distortion problem. We tackle this issue from two perspectives. Firstly, we introduce two distortion-adaptive modules. The first is a Distortion Mapping Module, which guides the model to pre-adapt to distorted features globally. The second module is a Distortion-Adaptive Attention Block that reduces local distortions on multi-scale features. Secondly, to exploit the unique characteristics of 360° data, we present a learnable relation matrix and use it as part of the positional embedding to further improve performance. Extensive experiments are conducted on three public datasets, and the results show that our model outperforms existing 2D SOD (salient object detection) and 360 SOD methods. The source code is available at https://github.com/yjzhao19981027/DATFormer/.

Label noise widely exists in large-scale image datasets. To mitigate the side effects of label noise, state-of-the-art methods focus on selecting confident examples by leveraging semi-supervised learning. Existing research shows that the ability to extract hard confident examples, which are close to the decision boundary, significantly influences the generalization ability of the learned classifier. In this paper, we find that a key reason for some hard examples being close to the decision boundary is due to the entanglement of style factors with content factors. The hard examples become more discriminative when we focus solely on content factors, such as semantic information, while ignoring style factors. Nonetheless, given only noisy data, content factors are not directly observed and have to be inferred. To tackle the problem of inferring content factors for classification when learning with noisy labels, our objective is to ensure that the content factors of all examples in the same underlying clean class remain unchanged as their style information changes. To achieve this, we utilize different data augmentation techniques to alter the styles while regularizing content factors based on some confident examples. By training existing methods with our inferred content factors, CS-Isolate proves their effectiveness in learning hard examples on benchmark datasets. The implementation is available at https://github.com/tmllab/2023_NeurIPS_CS-isolate.

Even though trained mainly on images, we discover that pretrained diffusion models show impressive power in guiding sketch synthesis. In this paper, we present DiffSketcher, an innovative algorithm that creates \textit{vectorized} free-hand sketches using natural language input. DiffSketcher is developed based on a pre-trained text-to-image diffusion model. It performs the task by directly optimizing a set of Bézier curves with an extended version of the score distillation sampling (SDS) loss, which allows us to use a raster-level diffusion model as a prior for optimizing a parametric vectorized sketch generator. Furthermore, we explore attention maps embedded in the diffusion model for effective stroke initialization to speed up the generation process. The generated sketches demonstrate multiple levels of abstraction while maintaining recognizability, underlying structure, and essential visual details of the subject drawn. Our experiments show that DiffSketcher achieves greater quality than prior work. The code and demo of DiffSketcher can be found at https://ximinng.github.io/DiffSketcher-project/.

Deep video compression is attracting increasing attention from both deep learning and video processing community. Recent learning-based approaches follow the hybrid coding paradigm to perform pixel space operations for reducing redundancy along both spatial and temporal dimentions, which leads to inaccurate motion estimation or less effective motion compensation. In this work, we propose a feature-space video coding framework (FVC), which performs all major operations (i.e., motion estimation, motion compression, motion compensation and residual compression) in the feature space. Specifically, a new deformable compensation module, which consists of motion estimation, motion compression and motion compensation, is proposed for more effective motion compensation. In our deformable compensation module, we first perform motion estimation in the feature space to produce the motion information (i.e., the offset maps). Then the motion information is compressed by using the auto-encoder style network. After that, we use the deformable convolution operation to generate the predicted feature for motion compensation. Finally, the residual information between the feature from the current frame and the predicted feature from the deformable compensation module is also compressed in the feature space. Motivated by the conventional codecs, in which the blocks with different sizes are used for motion estimation, we additionally propose two new modules called resolution-adaptive motion coding (RaMC) and resolution-adaptive residual coding (RaRC) to automatically cope with different types of motion and residual patterns at different spatial locations. Comprehensive experimental results demonstrate that our proposed framework achieves the state-of-the-art performance on three benchmark datasets including HEVC, UVG and MCL-JCV.

Although current salient object detection (SOD) works have achieved significant progress, they are limited when it comes to the integrity of the predicted salient regions. We define the concept of integrity at both a micro and macro level. Specifically, at the micro level, the model should highlight all parts that belong to a certain salient object. Meanwhile, at the macro level, the model needs to discover all salient objects in a given image. To facilitate integrity learning for SOD, we design a novel Integrity Cognition Network (ICON), which explores three important components for learning strong integrity features. 1) Unlike existing models, which focus more on feature discriminability, we introduce a diverse feature aggregation (DFA) component to aggregate features with various receptive fields (i.e., kernel shape and context) and increase feature diversity. Such diversity is the foundation for mining the integral salient objects. 2) Based on the DFA features, we introduce an integrity channel enhancement (ICE) component with the goal of enhancing feature channels that highlight the integral salient objects, while suppressing the other distracting ones. 3) After extracting the enhanced features, the part-whole verification (PWV) method is employed to determine whether the part and whole object features have strong agreement. Such part-whole agreements can further improve the micro-level integrity for each salient object. To demonstrate the effectiveness of our ICON, comprehensive experiments are conducted on seven challenging benchmarks. Our ICON outperforms the baseline methods in terms of a wide range of metrics. Notably, our ICON achieves ∼ 10% relative improvement over the previous best model in terms of average false negative ratio (FNR), on six datasets. Codes and results are available at: https://github.com/mczhuge/ICON.

Federated domain adaptation (FDA) is an effective method for performing learning tasks over distributed networks, which well improves data privacy and portability in unsupervised multi-source domain adaptation (UMDA) tasks. Despite the impressive gains achieved, two common limitations exist in current FDA works. First, most previous studies require access to the model parameters or gradient details of each source party. However, the raw source data can be reconstructed from the model gradients or parameters, which may leak individual information. Second, these works assume that different parties share an identical network architecture, which is impractical and not desirable for low- or high-resource target users. To address these issues, in this work, we propose a more practical UMDA setting, called Federated Multi-source Domain Adaptation on Black-box Models (B2FDA), where all data are stored locally and only the input-output interface of the source model is available. To tackle B2FDA, we propose an effective method, termed Co2-Learning with Multi-Domain Attention (Co-MDA). Experiments on multiple benchmark datasets demonstrate the effectiveness of our proposed method. Notably, Co-MDA performs comparably with traditional UMDA methods where the source data or the trained model are fully available.

Weakly supervised temporal action localization (WTAL) aims to localize actions in untrimmed videos with only weak supervision information (e.g., video-level labels). Most existing models handle all input videos with a fixed temporal scale. However, such models are not sensitive to actions whose pace of the movements is different from the “normal” speed, especially slow-motion action instances, which complete the movements with a much slower speed than their counterparts with a “normal” speed. Here arises the slow-motion blurred issue: It is hard to explore salient slow-motion information from videos at normal speed. In this paper, we propose a novel framework termed Slow Motion Enhanced Network (SMEN) to improve the ability of a WTAL network by compensating its sensitivity on slow-motion action segments. The proposed SMEN comprises a Mining module and a Localization module. The mining module generates mask to mine slow-motion-related features by utilizing the relationships between the normal motion and slow motion; while the localization module leverages the mined slow-motion features as complementary information to improve the temporal action localization results. Our proposed framework can be easily adapted by existing WTAL networks and enable them be more sensitive to slow-motion actions. Extensive experiments on three benchmarks are conducted, which demonstrate the high performance of our proposed framework.

In this paper, we propose MedoidsFormer, a novel transformer-based backbone equipped with a self-attention mechanism that is tailored explicitly to LiDAR-based 3D object detection. Unlike 2D object detection, the proportion of target objects to the input scene is much smaller, and their distribution is significantly sparser in 3D object detection. Given these observations, we introduce a new self-attention mechanism called Medoids Attention, focusing on exploiting interactions within surrounding regions, which not only reduces computation and memory costs but obtains discriminative context information. Instead of aggregating tokens from adjacent areas, we present a dynamic semantic-aware token mining process through k-Medoids clustering to direct select representative tokens for attention modeling. Our proposed method shows consistent improvement over existing 3D object detectors through extensive experiments and achieves state-of-the-art performance on the large-scale Waymo Open Dataset. We also conduct comprehensive ablation studies to verify the efficacy of the new self-attention mechanism and provide thorough insights.

Point cloud recognition has recently gained increasing research interest due to the huge potential in real-world applications such as autonomous driving, robotics, etc. However, the point clouds of similar objects often exhibit notable geometric variations due to the difference in capturing devices or environmental changes. This leads to significant performance degradation when the learned point cloud recognition model is applied to a new scenario, which is also known as the domain adaptation issue. In this work, we propose a new unsupervised domain adaptation approach for point cloud recognition via domain adaptive sampling (DAS). In particular, we propose a two-level sampling strategy of point level and instance level to improve the cross-domain recognition ability of the model. First, we propose a domain adaptive point sampling (DAPS) strategy to enhance the domain-invariant representation of point clouds by progressively focusing on representative points in each point cloud based on geometric consistency. Then, we further propose an instance-level domain adaptive cloud sampling (DACS) strategy to learn target-specific information based on a self-paced learning paradigm, where we select a set of pseudo-labeled target point clouds to train our designed light-weighted adapters without modifying the learned domain-invariant representation. We validate our domain adaptive sampling approach on the benchmark datasets PointDA-10 and GraspNetPC-10, where our method achieves new state-of-the-art performance.

Recently, 3D vision-and-language tasks have attracted increasing research interest. Compared to other vision-and-language tasks, the 3D visual question answering (VQA) task is less exploited and is more susceptible to language priors and co-reference ambiguity. Meanwhile, a couple of recently proposed 3D VQA datasets do not well support 3D VQA task due to their limited scale and annotation methods. In this work, we formally define and address a 3D grounded question answering (GQA) task by collecting a new 3D VQA dataset, referred to as flexible and explainable 3D GQA (FE-3DGQA), with diverse and relatively free-form question-answer pairs, as well as dense and completely grounded bounding box annotations. To achieve more explainable answers, we label the objects appeared in the complex QA pairs with different semantic types, including answer-grounded objects (both appeared and not appeared in the questions), and contextual objects for answer-grounded objects. We also propose a new 3D VQA framework to effectively predict the completely visually grounded and explainable answer. Extensive experiments verify that our newly collected benchmark datasets can be effectively used to evaluate various 3D VQA methods from different aspects and our newly proposed framework also achieves the state-of-the-art performance on the new benchmark dataset. The datasets and the source code are available viahttps://github.com/zlccccc/3DVL_Codebase.

In this work, we propose a new deep image compression framework called Complexity and Bitrate Adaptive Network (CBANet) that aims to learn one single network to support variable bitrate coding under various computational complexity levels. In contrast to the existing state-of-the-art learning-based image compression frameworks that only consider the rate-distortion trade-off without introducing any constraint related to the computational complexity, our CBANet considers the complex rate-distortion-complexity trade-off when learning a single network to support multiple computational complexity levels and variable bitrates. Since it is a non-trivial task to solve such a rate-distortion-complexity related optimization problem, we propose a two-step approach to decouple this complex optimization task into a complexity-distortion optimization sub-task and a rate-distortion optimization sub-task, and additionally propose a new network design strategy by introducing a Complexity Adaptive Module (CAM) and a Bitrate Adaptive Module (BAM) to respectively achieve the complexity-distortion and rate-distortion trade-offs. As a general approach, our network design strategy can be readily incorporated into different deep image compression methods to achieve complexity and bitrate adaptive image compression by using a single network. Comprehensive experiments on two benchmark datasets demonstrate the effectiveness of our CBANet for deep image compression. Code is released at https://github.com/JinyangGuo/CBANet-release.

Deep learning based image enhancement models have largely improved the readability of fundus images in order to decrease the uncertainty of clinical observations and the risk of misdiagnosis. However, due to the difficulty of acquiring paired real fundus images at different qualities, most existing methods have to adopt synthetic image pairs as training data. The domain shift between the synthetic and the real images inevitably hinders the generalization of such models on clinical data. In this work, we propose an end-to-end optimized teacher-student framework to simultaneously conduct image enhancement and domain adaptation. The student network uses synthetic pairs for supervised enhancement, and regularizes the enhancement model to reduce domain-shift by enforcing teacher-student prediction consistency on the real fundus images without relying on enhanced ground-truth. Moreover, we also propose a novel multi-stage multi-attention guided enhancement network (MAGE-Net) as the backbones of our teacher and student network. Our MAGE-Net utilizes multi-stage enhancement module and retinal structure preservation module to progressively integrate the multi-scale features and simultaneously preserve the retinal structures for better fundus image quality enhancement. Comprehensive experiments on both real and synthetic datasets demonstrate that our framework outperforms the baseline approaches. Moreover, our method also benefits the downstream clinical tasks.

In this work, we propose a new end-to-end optimized two-stream framework called GeometryMotion-Transformer (GMT) for 3D action recognition. We first observe that the existing 3D action recognition approaches cannot well extract motion representations from point cloud sequences. Specifically, when extracting motion representations, the existing approaches do not explicitly consider one-to-one correspondence among frames. Besides, the existing methods only extract thesingle-scalemotion representations, which cannot well model the complex motion patterns of moving objects in point cloud sequences. To address these issues, we first propose the feature extraction module (FEM) to generate one-to-one correspondence among frames without using the voxelization process, and explicitly extract both geometry andmulti-scalemotion representations from raw point clouds. Moreover, we also observe the existing two-stream 3D action recognition approaches simply concatenate or add the geometry and motion features, which cannot well exploit the relationship between two-steam features. To this end, we also propose an improved transformer-based feature fusion module (FFM) to effectively fuse the two-stream features. Based on the proposed FEM and FFM, we build our GMT for 3D action recognition. Extensive experimental results on four benchmark datasets demonstrate the effectiveness of our backbone GMT.

Observing that the 3D captioning task and the 3D grounding task contain both shared and complementary information in nature, in this work, we propose a unified framework to jointly solve these two distinct but closely related tasks in a synergistic fashion, which consists of both shared task-agnostic modules and lightweight task-specific modules. On one hand, the shared task-agnostic modules aim to learn precise locations of objects, fine-grained attribute features to characterize different objects, and complex relations between objects, which benefit both captioning and visual grounding. On the other hand, by casting each of the two tasks as the proxy task of another one, the lightweight task-specific modules solve the captioning task and the grounding task respectively. Extensive experiments and ablation study on three 3D vision and language datasets demonstrate that our joint training frame-work achieves significant performance gains for each individual task and finally improves the state-of-the-art performance for both captioning and grounding tasks.

In this work, we propose the first end-to-end optimized framework for compressing automotive stereo videos (i.e., stereo videos from autonomous driving applications) from both left and right views. Specifically, when compressing the current frame from each view, our framework reduces temporal redundancy by performing motion compensation using the reconstructed intra-view adjacent frame and at the same time exploits binocular redundancy by conducting disparity compensation using the latest reconstructed cross-view frame. Moreover, to effectively compress the introduced motion and disparity offsets for better compensation, we further propose two novel schemes called motion residual compression and disparity residual compression to respectively generate the predicted motion offset and disparity offset from the previously compressed motion offset and disparity offset, such that we can more effectively compress residual offset information for better bit-rate saving. Overall, the entire framework is implemented by the fully-differentiable modules and can be optimized in an end-to-end manner. Our comprehensive experiments on three automotive stereo video benchmarks Cityscapes, KITTI 2012 and KITTI 2015 demonstrate that our proposed framework outperforms the learning-based single-view video codec and the traditional hand-crafted multi-view video codec.

The previous deep video compression approaches only use the single scale motion compensation strategy and rarely adopt the mode prediction technique from the traditional standards like H.264/H.265 for both motion and residual compression. In this work, we first propose a coarse-to-fine (C2F) deep video compression framework for better motion compensation, in which we perform motion estimation, compression and compensation twice in a coarse to fine manner. Our C2F framework can achieve better motion compensation results without significantly increasing bit costs. Observing hyperprior information (i.e., the mean and variance values) from the hyperprior networks contains discriminant statistical information of different patches, we also propose two efficient hyperprior-guided mode prediction methods. Specifically, using hyper-prior information as the input, we propose two mode prediction networks to respectively predict the optimal block resolutions for better motion coding and decide whether to skip residual information from each block for better residual coding without introducing additional bit cost while bringing negligible extra computation cost. Comprehensive experimental results demonstrate our proposed C2F video compression framework equipped with the new hyperprior-guided mode prediction methods achieves the state-of-the-art performance on HEVC, UVG and MCL-JCV datasets.

Multi-modality (i.e., multi-sensor) data is widely used in various vision tasks for more accurate or robust perception. However, the increased data modalities bring new challenges for data storage and transmission. The existing data compression approaches usually adopt individual codecs for each modality without considering the correlation between different modalities. This work proposes a multi-modality compression framework for infrared and visible image pairs by exploiting the cross-modality redun-dancy. Specifically, given the image in the reference modality (e.g., the infrared image), we use the channel-wise alignment module to produce the aligned features based on the affine transform. Then the aligned feature is used as the context information for compressing the image in the current modality (e.g., the visible image), and the corresponding affine coefficients are losslessly compressed at negligible cost. Furthermore, we introduce the Transformer-based spatial alignment module to exploit the correlation between the intermediate features in the decoding procedures for different modalities. Our framework is very flexible and easily extended for multi-modality video compression. Experimental results show our proposed framework outperforms the traditional and learning-based single modality compression methods on the FLIR and KAIST datasets.

This paper proposes an end-to-end fine-grained visual categorization system, termed Part-based Convolutional Neural Network (P-CNN), which consists of three modules. The first module is a Squeeze-and-Excitation (SE) block, which learns to recalibrate channel-wise feature responses by emphasizing informative channels and suppressing less useful ones. The second module is a Part Localization Network (PLN) used to locate distinctive object parts, through which a bank of convolutional filters are learned as discriminative part detectors. Thus, a group of informative parts can be discovered by convolving the feature maps with each part detector. The third module is a Part Classification Network (PCN) that has two streams. The first stream classifies each individual object part into image-level categories. The second stream concatenates part features and global feature into a joint feature for the final classification. In order to learn powerful part features and boost the joint feature capability, we propose a Duplex Focal Loss used for metric learning and part classification, which focuses on training hard examples. We further merge PLN and PCN into a unified network for an end-to-end training process via a simple training technique. Comprehensive experiments and comparisons with state-of-the-art methods on three benchmark datasets demonstrate the effectiveness of our proposed method.

Deep neural networks designed for point clouds, also called point cloud neural networks (PCNNs), are attracting increasing attention in recent years. In this work, we propose the first model compression framework referred to as JointPruning (JP) that is specifically designed for compressing PCNNs. Observing that the redundancies in PCNNs are largely affected by certain parameters like the number of points, we first propose a new search space specifically designed for PCNNs. By searching the optimal pruning policy in our newly proposed search space, our JP framework can prune the PCNNs at different levels and simultaneously reduce the redundancies along multiple dimensions. As the newly proposed search space consists of multiple levels and the policy value at each level is continuous in our JP framework, it is hard to directly search for the best pruning policy in such a large search space. To this end, we further propose two strategies called search space refinement and validation set extension to progressively refine the granularity of our searching process in a coarse-to-fine and easy-to-hard fashion, which can help us gradually find better pruning policies. Comprehensive experiments on three benchmark datasets demonstrate the effectiveness of our JP framework for compressing PCNNs.

The existing end-to-end optimized 3D action recognition methods often suffer from high computational costs. Observing that different frames and different points in point cloud sequences often have different importance values for the 3D action recognition task, in this work, we propose a fully automatic model compression framework called 3D-Pruning (3DP) for efficient 3D action recognition. After performing model compression by using our 3DP framework, the compressed model can process different frames and different points in each frame by using different computational complexities based on their importance values, in which both the importance value and computational complexity for each frame/point can be automatically learned. Extensive experiments on five benchmark datasets demonstrate the effectiveness of our 3DP framework for model compression.

In this work, we introduce a novel task – Human-centric Spatio-Temporal Video Grounding (HC-STVG). Unlike the existing referring expression tasks in images or videos, by focusing on humans, HC-STVG aims to localize a spatio-temporal tube of the target person from an untrimmed video based on a given textural description. This task is useful, especially for healthcare and security related applications, where the surveillance videos can be extremely long but only a specific person during a specific period is concerned. HC-STVG is a video grounding task that requires both spatial (where) and temporal (when) localization. Unfortunately, the existing grounding methods cannot handle this task well. We tackle this task by proposing an effective baseline method named Spatio-Temporal Grounding with Visual Transformers (STGVT), which utilizes Visual Transformers to extract cross-modal representations for video-sentence matching and temporal localization. To facilitate this task, we also contribute an HC-STVG datasetThe new dataset is available athttps://github.com/tzhhhh123/HC-STVG. consisting of 5,660 video-sentence pairs on complex multi-person scenes. Specifically, each video lasts for 20 seconds, pairing with a natural query sentence with an average of 17.25 words. Extensive experiments are conducted on this dataset, demonstrating that the newly-proposed method outperforms the existing baseline methods.

Domain adaptation aims to leverage a label-rich domain (the source domain) to help model learning in a label-scarce domain (the target domain). Most domain adaptation methods require the co-existence of source and target domain samples to reduce the distribution mismatch. However, access to the source domain samples may not always be feasible in real-world applications due to different problems (e.g., storage, transmission, and privacy issues). In this work, we deal with the source data-free unsupervised domain adaptation problem and propose a novel approach referred to as Virtual Domain Modeling for Domain Adaptation (VDM-DA), in which the virtual domain acts as a bridge between the source and target domains. Specifically, based on the pre-trained source model, we generate the virtual domain samples by using an approximated Gaussian Mixture Model (GMM) in the feature space, such that the virtual domain maintains a similar distribution with the source domain without access to the original source data. Moreover, we also design an effective distribution alignment method to reduce the distribution divergence between the virtual domain and the target domain by gradually improving the compactness of the target domain distribution through model learning. In this way, we successfully achieve the goal of distribution alignment between the source and target domains when training deep networks without access to the source domain data. We conduct extensive experiments on four benchmark datasets for both 2D image-based and 3D point cloud-based cross-domain object recognition tasks, where the proposed method referred to as Virtual Domain Modeling for Domain Adaptation (VDM-DA) achieves the promising performance on all datasets.

3D volumetric image processing has attracted increasing attention in the last decades, in which one major research area is to develop efficient lossless volumetric image compression techniques to better store and transmit such images with massive amount of information. In this work, we propose the first end-to-end optimized learning framework for losslessly compressing 3D volumetric data. Our approach builds upon a hierarchical compression scheme by additionally introducing the intra-slice auxiliary features and estimating the entropy model based on both intra-slice and inter-slice latent priors. Specifically, we first extract the hierarchical intra-slice auxiliary features through multi-scale feature extraction modules. Then, an Intra-slice and Inter-slice Conditional Entropy Coding module is proposed to fuse the intra-slice and inter-slice information from different scales as the context information. Based on such context information, we can predict the distributions for both intra-slice auxiliary features and the slice images. To further improve the lossless compression performance, we also introduce two new gating mechanisms called Intra-Gate and Inter-Gate to generate the optimal feature representations for better information fusion. Eventually, we can produce the bitstream for losslessly compressing volumetric images based on the estimated entropy model. Different from the existing lossless volumetric image codecs, our end-to-end optimized framework jointly learns both intra-slice auxiliary features at different scales for each slice and inter-slice latent features from previously encoded slices for better entropy estimation. The extensive experimental results indicate that our framework outperforms the state-of-the-art hand-crafted lossless volumetric image codecs (e.g., JP3D) and the learning-based lossless image compression method on four volumetric image benchmarks for losslessly compressing both 3D Medical Images and Hyper-Spectral Images.

Unsupervised active learning has become an active research topic in the machine learning and computer vision communities, whose goal is to choose a subset of representative samples to be labeled in an unsupervised setting. Most of existing approaches rely on shallow linear models by assuming that each sample can be well approximated by the span (i.e., the set of all linear combinations) of the selected samples, and then take these selected samples as the representative ones for manual labeling. However, the data do not necessarily conform to the linear models in many real-world scenarios, and how to model nonlinearity of data often becomes the key point of unsupervised active learning. Moreover, the existing works often aim to well reconstruct the whole dataset, while ignore the important cluster structure, especially for imbalanced data. In this paper, we present a novel deep unsupervised active learning framework. The proposed method can explicitly learn a nonlinear embedding to map each input into a latent space via a deep neural network, and introduce a selection block to select the representative samples in the learnt latent space through a self-supervised learning strategy. In the selection block, we aim to not only preserve the global structure of the data, but also capture the cluster structure of the data in order to well handle the data imbalance issue during sample selection. Meanwhile, we take advantage of the clustering result to provide self-supervised information to guide the above processes. Finally, we attempt to preserve the local structure of the data, such that the data embedding becomes more precise and the model performance can be further improved. Extensive experimental results on several publicly available datasets clearly demonstrate the effectiveness of our method, compared with the state-of-the-arts.

Observing that it is still a challenging task to deploy 3D action recognition methods in real-world scenarios, in this work, we investigate the accuracy-efficiency trade-off for 3D action recognition. We first introduce a simple and efficient backbone network structure for 3D action recognition, in which we directly extract the geometry and motion representations from the raw point cloud videos through a set of simple operations (i.e., coordinate offset generation and mini-PoinNet). Based on the backbone network, we propose an end-to-end optimized network called adaptive point sampling network (APSNet) to achieve the accuracy-efficiency trade-off, which mainly consists of three stages: the coarse feature extraction stage, the decision making stage, and the fine feature extraction stage. In APSNet, we adaptively decide the optimal resolutions (i.e., the optimal number of points) for each pair of frames based on any input point cloud video under the given computational complexity constraint. Comprehensive experiments on multiple benchmark datasets demonstrate the effectiveness and efficiency of our newly proposed APSNet for 3D action recognition.

In this work, we propose a new patch-based framework called VPU for the video-based point cloud upsampling task by effectively exploiting temporal dependency among multiple consecutive point cloud frames, in which each frame consists of a set of unordered, sparse and irregular 3D points. Rather than adopting the sophisticated motion estimation strategy in video analysis, we propose a new spatio-temporal aggregation (STA) module to effectively extract, align and aggregate rich local geometric clues from consecutive frames at the feature level. By more reliably summarizing spatio-temporally consistent and complementary knowledge from multiple frames in the resultant local structural features, our method better infers the local geometry distributions at the current frame. In addition, our STA module can be readily incorporated with various existing single frame-based point upsampling methods (e.g., PU-Net, MPU, PU-GAN and PU-GCN). Comprehensive experiments on multiple point cloud sequence datasets demonstrate our video-based point cloud upsampling framework achieves substantial performance improvement over its single frame-based counterparts.

3D object detection in point clouds is a challenging vision task that benefits various applications for understanding the 3D visual world. Lots of recent research focuses on how to exploit end-to-end trainable Hough voting for generating object proposals. However, the current voting strategy can only receive partial votes from the surfaces of potential objects together with severe outlier votes from the cluttered backgrounds, which hampers full utilization of the information from the input point clouds. Inspired by the back-tracing strategy in the conventional Hough voting methods, in this work, we introduce a new 3D object detection method, named as Back-tracing Representative Points Network (BRNet), which generatively back-traces the representative points from the vote centers and also revisits complementary seed points around these generated points, so as to better capture the fine local structural features surrounding the potential objects from the raw point clouds. Therefore, this bottom-up and then top-down strategy in our BRNet enforces mutual consistency between the predicted vote centers and the raw surface points and thus achieves more reliable and flexible object localization and class prediction results. Our BRNet is simple but effective, which significantly outperforms the state-of-the-art methods on two large-scale point cloud datasets, ScanNet V2 (+7.5% in terms of [email protected]) and SUN RGB-D (+4.7% in terms of [email protected]), while it is still lightweight and efficient.

Learning based video compression attracts increasing attention in the past few years. The previous hybrid coding approaches rely on pixel space operations to reduce spatial and temporal redundancy, which may suffer from inaccurate motion estimation or less effective motion compensation. In this work, we propose a feature-space video coding network (FVC) by performing all major operations (i.e., motion estimation, motion compression, motion compensation and residual compression) in the feature space. Specifically, in the proposed deformable compensation module, we first apply motion estimation in the feature space to produce motion information (i.e., the offset maps), which will be compressed by using the auto-encoder style network. Then we perform motion compensation by using deformable convolution and generate the predicted feature. After that, we compress the residual feature between the feature from the current frame and the predicted feature from our deformable compensation module. For better frame reconstruction, the reference features from multiple previous reconstructed frames are also fused by using the nonlocal attention mechanism in the multi-frame feature fusion module. Comprehensive experimental results demonstrate that the proposed framework achieves the state-of-the-art performance on four benchmark datasets including HEVC, UVG, VTL and MCL-JCV.

As a variant of standard convolution, a dilated convolution can control effective receptive fields and handle large scale variance of objects without introducing additional computational costs. To fully explore the potential of dilated convolution, we proposed a new type of dilated convolution (referred to as inception convolution), where the convolution operations have independent dilation patterns among different axes, channels and layers. To develop a practical method for learning complex inception convolution based on the data, a simple but effective search algorithm, referred to as efficient dilation optimization (EDO), is developed. Based on statistical optimization, the EDO method operates in a low-cost manner and is extremely fast when it is applied on large scale datasets. Empirical results validate that our method achieves consistent performance gains for image recognition, object detection, instance segmentation, human detection, and human pose estimation. For instance, by simply replacing the 3 × 3 standard convolution in the ResNet-50 backbone with inception convolution, we significantly improve the AP of Faster R-CNN from 36.4% to 39.2% on MS COCO.

In this paper, we propose a two-stage deep learning framework called VoxelContext-Net for both static and dynamic point cloud compression. Taking advantages of both octree based methods and voxel based schemes, our approach employs the voxel context to compress the octree structured data. Specifically, we first extract the local voxel representation that encodes the spatial neighbouring context information for each node in the constructed octree. Then, in the entropy coding stage, we propose a voxel context based deep entropy model to compress the symbols of non-leaf nodes in a lossless way. Furthermore, for dynamic point cloud compression, we additionally introduce the local voxel representations from the temporal neighbouring point clouds to exploit temporal dependency. More importantly, to alleviate the distortion from the octree construction procedure, we propose a voxel context based 3D coordinate refinement method to produce more accurate reconstructed point cloud at the decoder side, which is applicable to both static and dynamic point cloud compression. The comprehensive experiments on both static and dynamic point cloud benchmark datasets(e.g., ScanNet and Semantic KITTI) clearly demonstrate the effectiveness of our newly proposed method VoxelContext-Net for 3D point cloud geometry compression.

Geometric characteristic plays an important role in the representation of an object in 3D point clouds. For example, large objects often contain more points, while small ones contain fewer points. The points from objects near the capture device are denser, while those from far-range objects are sparser. These issues bring new challenges to 3D object detection, especially under the domain adaptation scenarios. In this work, we propose a new cross-dataset 3D object detection method named Scale-aware and Range-aware Domain Adaptation Network (SRDAN). We take advantage of the geometric characteristics of 3D data (i.e., size and distance), and propose the scale-aware domain alignment and the range-aware domain alignment strategies to guide the distribution alignment between two domains. For scale-aware domain alignment, we design a 3D voxel-based feature pyramid network to extract multi-scale semantic voxel features, and align the features and instances with similar scales between two domains. For range-aware domain alignment, we introduce a range-guided domain alignment module to align the features of objects according to their distance to the capture device. Extensive experiments under three different scenarios demonstrate the effectiveness of our SRDAN approach, and comprehensive ablation study also validates the importance of geometric characteristics for cross-dataset 3D object detection.

Spatio-temporal video grounding (STVG) aims to localize a spatio-temporal tube of a target object in an untrimmed video based on a query sentence. In this work, we propose a one-stage visual-linguistic transformer based framework called STVGBert for the STVG task, which can simultaneously localize the target object in both spatial and temporal domains. Specifically, without resorting to pregenerated object proposals, our STVGBert directly takes a video and a query sentence as the input, and then produces the cross-modal features by using the newly introduced cross-modal feature learning module ST-ViLBert. Based on the cross-modal features, our method then generates bounding boxes and predicts the starting and ending frames to produce the predicted object tube. To the best of our knowledge, our STVGBert is the first one-stage method, which can handle the STVG task without relying on any pre-trained object detectors. Comprehensive experiments demonstrate our newly proposed framework outperforms the state-of-the-art multi-stage methods on two benchmark datasets Vid-STG and HC-STVG.

In this work, we propose a new feed-forward arbitrary style transfer method, referred to as StyleFormer, which can simultaneously fulfill fine-grained style diversity and semantic content coherency. Specifically, our transformer-inspired feature-level stylization method consists of three modules: (a) the style bank generation module for sparse but compact parametric style pattern extraction, (b) the transformer-driven style composition module for content-guided global style composition, and (c) the parametric content modulation module for flexible but faithful stylization. The output stylized images are impressively coherent with the content structure, sensitive to the detailed style variations, but still holistically adhere to the style distributions from the style images. Qualitative and quantitative comparisons as well as comprehensive user studies demonstrate that our StyleFormer outperforms the existing SOTA methods in generating visually plausible stylization results with real-time efficiency.

Visual grounding on 3D point clouds is an emerging vision and language task that benefits various applications in understanding the 3D visual world. By formulating this task as a grounding-by-detection problem, lots of recent works focus on how to exploit more powerful detectors and comprehensive language features, but (1) how to model complex relations for generating context-aware object proposals and (2) how to leverage proposal relations to distinguish the true target object from similar proposals are not fully studied yet. Inspired by the well-known transformer architecture, we propose a relation-aware visual grounding method on 3D point clouds, named as 3DVG-Transformer, to fully utilize the contextual clues for relation-enhanced proposal generation and cross-modal proposal disambiguation, which are enabled by a newly designed coordinate-guided contextual aggregation (CCA) module in the object proposal generation stage, and a multiplex attention (MA) module in the cross-modal feature fusion stage. We validate that our 3DVG-Transformer outperforms the state-of-the-art methods by a large margin, on two point cloud-based visual grounding datasets, ScanRefer and Nr3D/Sr3D from ReferIt3D, especially for complex scenarios containing multiple objects of the same category.

Automatic check-out (ACO) emerges as an integral component in recent self-service retailing stores, which aims at automatically detecting and counting the randomly placed products upon a check-out platform. Existing data-driven counting works still have difficulties in generalizing to real-world retail product counting scenarios, since (1) real check-out images are hard to collect or cover all products and their possible layouts, (2) rapid updating of the product list leads to frequent and tedious re-training of the counting models. To overcome these obstacles, we contribute a practical automatic check-out framework tailored to real-world retail product counting scenarios, consisting of a photorealistic exemplar augmentation to generate physically reliable and photorealistic check-out images from canonical exemplars scanned for each product and an incremental learning strategy to match the updating nature of the ACO system with much fewer training effort. Through comprehensive studies, we show that the proposed IncreACO serves as an effective framework on the recent Retail Product Checkout (RPC) dataset, where the proposed photorealistic exemplar augmentation remarkably improves the counting performance against the state-of-the-art methods (77.15% v.s. 72.83% in counting accuracy), whilst the proposed incremental learning framework consistently extends the counting performance to new categories.

Category-specific 3D object shape models have greatly boosted the recent advances in object detection, recognition and segmentation. However, even the most advanced approach for learning 3D object shapes still requires heavy manual annotations on large-scale 2D images. Such annotations include object categories, object keypoints, and figure-ground segmentation for the instances in each image. In particular, annotating figure-ground segmentation is unbearably labor-intensive and time-consuming. To address this problem, this paper devotes to learn category-specific 3D shape models under weak supervision, where only object categories and keypoints are required to be manually annotated on the training 2D images. By exploring the underlying relationship between two tasks: object segmentation and category-specific 3D shape reconstruction, we propose a novel weakly-supervised learning framework to jointly address these two tasks and combine them to boost the final performance of the learned 3D shape models. Moreover, learning without using figure-ground segmentation leads to ambiguous solutions. To this end, we develop the confidence weighting schemes in the viewpoint estimation and 3D shape learning procedure. These schemes effectively reduce the confusion caused by the noisy data and thus increase the chances for recovering more reliable 3D object shapes. Comprehensive experiments on the challenging PASCAL VOC benchmark show that our framework achieves comparable performance with the state-of-the-art methods that use expensive manual segmentation-level annotations. In addition, our experiments also demonstrate that our 3D shape models improve object segmentation performance.

Traditional video compression approaches build upon the hybrid coding framework with motion-compensated prediction and residual transform coding. In this paper, we propose the first end-to-end deep video compression framework to take advantage of both the classical compression architecture and the powerful non-linear representation ability of neural networks. Our framework employs pixel-wise motion information, which is learned from an optical flow network and further compressed by an auto-encoder network to save bits. The other compression components are also implemented by the well-designed networks for high efficiency. All the modules are jointly optimized by using the rate-distortion trade-off and can collaborate with each other. More importantly, the proposed deep video compression framework is very flexible and can be easily extended by using lightweight or advanced networks for higher speed or better efficiency. We also propose to introduce the adaptive quantization layer to reduce the number of parameters for variable bitrate coding. Comprehensive experimental results demonstrate the effectiveness of the proposed framework on the benchmark datasets.

Spatio-temporal action localization consists of three levels of tasks: spatial localization, action classification, and temporal localization. In this work, we propose a new progressive cross-stream cooperation (PCSC) framework that improves all three tasks above. The basic idea is to utilize both spatial region (resp., temporal segment proposals) and features from one stream (i.e., the Flow/RGB stream) to help another stream (i.e., the RGB/Flow stream) to iteratively generate better bounding boxes in the spatial domain (resp., temporal segments in the temporal domain). In this way, not only the actions could be more accurately localized both spatially and temporally, but also the action classes could be predicted more precisely. Specifically, we first combine the latest region proposals (for spatial detection) or segment proposals (for temporal localization) from both streams to form a larger set of labelled training samples to help learn better action detection or segment detection models. Second, to learn better representations, we also propose a new message passing approach to pass information from one stream to another stream, which also leads to better action detection and segment detection models. By first using our newly proposed PCSC framework for spatial localization at the frame-level and then applying our temporal PCSC framework for temporal localization at the tube-level, the action localization results are progressively improved at both the frame level and the video level. Comprehensive experiments on two benchmark datasets UCF-101-24 and J-HMDB demonstrate the effectiveness of our newly proposed approaches for spatio-temporal action localization in realistic scenarios.

This paper proposes a new unsupervised domain adaptation approach called Collaborative and Adversarial Network (CAN), which uses the domain-collaborative and domain-adversarial learning strategies for training the neural network. The domain-collaborative learning strategy aims to learn domain specific feature representation to preserve the discriminability for the target domain, while the domain adversarial learning strategy aims to learn domain invariant feature representation to reduce the domain distribution mismatch between the source and target domains. We show that these two learning strategies can be uniformly formulated as domain classifier learning with positive or negative weights on the losses. We then design a collaborative and adversarial training scheme, which automatically learns domain specific representations from lower blocks in CNNs through collaborative learning and domain invariant representations from higher blocks through adversarial learning. Moreover, to further enhance the discriminability in the target domain, we propose Self-Paced CAN (SPCAN), which progressively selects pseudo-labeled target samples for re-training the classifiers. We employ a self-paced learning strategy such that we can select pseudo-labeled target samples in an easy-to-hard fashion. Additionally, we build upon the popular two-stream approach to extend our domain adaptation approach for more challenging video action recognition task, which additionally considers the cooperation between the RGB stream and the optical flow stream. We propose the Two-stream SPCAN (TS-SPCAN) method to select and reweight the pseudo labeled target samples of one stream (RGB/Flow) based on the information from the other stream (Flow/RGB) in a cooperative way. As a result, our TS-SPCAN model is able to exchange the information between the two streams. Comprehensive experiments on different benchmark datasets, Office-31, ImageCLEF-DA and VISDA-2017 for the object recognition task, and UCF101-10 and HMDB51-10 for the video action recognition task, show our newly proposed approaches achieve the state-of-the-art performance, which clearly demonstrates the effectiveness of our proposed approaches for unsupervised domain adaptation.

In this work, we propose a simple but effective channel pruning framework called Progressive Channel Pruning (PCP) to accelerate Convolutional Neural Networks (CNNs). In contrast to the existing channel pruning methods that prune the channels only once per layer in a layer-by-layer fashion, our new progressive framework iteratively prunes a small number of channels from several selected layers, which consists of a three-step attempting-selecting-pruning pipeline in each iteration. In the attempting step, we attempt to prune a pre-defined number of channels from one layer by using any existing channel pruning methods and estimate the accuracy drop for this layer based on the labelled samples in the validation set. In the selecting step, based on the estimated accuracy drops for all layers, we propose a greedy strategy to automatically select a set of layers that will lead to less overall accuracy drop after pruning these layers. In the pruning step, we prune a small number of channels from these selected layers. We further extend our PCP framework to prune channels for the deep transfer learning methods like Domain Adversarial Neural Network (DANN), in which we effectively reduce the data distribution mismatch in the channel pruning process by using both labelled samples from the source domain and pseudo-labelled samples from the target domain. Our comprehensive experiments on two benchmark datasets demonstrate that our PCP framework outperforms the existing channel pruning approaches under both supervised learning and transfer learning settings.

In this work, we propose a strong two-stream baseline method referred to as GeometryMotion-Net for 3D action recognition. For efficient 3D action recognition, we first represent each point cloud sequence as a limited number of randomly sampled frames with each frame consisting of a sparse set of points. After that, we propose a new two-stream framework for effective 3D action recognition. For the geometry stream, we propose a new module to produce a virtual overall geometry point cloud by first merging all 3D points from these selected frames, and then we exploit local neighborhood information of each point in the feature space. In the motion stream, for any two neighboring point cloud frames, we also propose a new module to generate one virtual forward motion point cloud and one virtual backward motion point cloud. Specifically, for each point in the current frame, we first produce a set of 3D offset features relative to the neighboring points in the reference frame (i.e., the previous/subsequent frame) and then exploit local neighborhood information of this point in the offset feature space. Based on the newly generated virtual overall geometry point cloud and multiple virtual forward/backward motion point clouds, any existing point cloud analysis methods (e.g., PointNet) can be readily adopted for extracting discriminant geometry and bidirectional motion features in the geometry and motion streams, respectively, which are further aggregated to make our two-stream network trainable in an end-to-end fashion. Comprehensive experiments on both large-scale datasets (i.e.NTU RGB+D 60 and NTU RGB+D 120) and small-scale datasets (i.e., N-UCLA and UWA3DII) demonstrate the effectiveness and efficiency of our two-stream network for 3D action recognition.

In this work, we propose a new sequential point cloud upsampling method called SPU, which aims to upsample sparse, non-uniform, and orderless point cloud sequences by effectively exploiting rich and complementary temporal dependency from multiple inputs. Specifically, these inputs include a set of multi-scale short-term features from the 3D points in three consecutive frames (i.e., the previous/current/subsequent frame) and a long-term latent representation accumulated throughout the point cloud sequence. Considering that these temporal clues are not well aligned in the coordinate space, we propose a new temporal alignment module (TAM) based on the cross-attention mechanism to transform each individual feature into the feature space of the current frame. We also propose a new gating mechanism to learn the optimal weights for these transformed features, based on which the transformed features can be effectively aggregated as the final fused feature. The fused feature can be readily fed into the existing single frame-based point cloud upsampling methods (e.g., PU-Net, MPU and PU-GAN) to generate the dense point cloud for the current frame. Comprehensive experiments on three benchmark datasets DYNA, COMA, and MSR Action3D demonstrate the effectiveness of our method for upsampling point cloud sequences.

Voting-based methods (e.g., VoteNet) have achieved promising results for 3D object detection. However, the simple voting operation in VoteNet may lead to less accurate voting results that are far away from the true object centers. In this work, we propose a simple but effective 3D object detection method called Transformer3D-Det (T3D), in which we additionally introduce a transformer based vote refinement module to refine the voting results of VoteNet and can thus significantly improve the 3D object detection performance. Specifically, our T3D framework consists of three modules: a vote generation module, a vote refinement module, and a bounding box generation module. Given an input point cloud, we first utilize the vote generation module to generate multiple coarse vote clusters. Then, the clustered coarse votes will be refined by using our transformer based vote refinement module to produce more accurate and meaningful votes. Finally, the bounding box generation module takes the refined vote clusters as the input and generates the final detection result for the input point cloud. To alleviate the impact of inaccurate votes, we also propose a new non-vote loss function to train our T3D. As a result, our T3D framework can achieve better 3D object detection performance. Comprehensive experiments on two benchmark datasets ScanNetV2 and SUN RGB-D demonstrate the effectiveness of our T3D framework for 3D object detection.

Most existing unsupervised active learning methods aim at minimizing the data reconstruction loss by using the linear models to choose representative samples for manually labeling in an unsupervised setting. Thus these methods often fail in modelling data with complex non-linear structure. To address this issue, we propose a new deep unsupervised Active Learning method for classification tasks, inspired by the idea of Matrix Sketching, called ALMS. Specifically, ALMS leverages a deep auto-encoder to embed data into a latent space, and then describes all the embedded data with a small size sketch to summarize the major characteristics of the data. In contrast to previous approaches that reconstruct the whole data matrix for selecting the representative samples, ALMS aims to select a representative subset of samples to well approximate the sketch, which can preserve the major information of data meanwhile significantly reducing the number of network parameters. This makes our algorithm alleviate the issue of model overfitting and readily cope with large datasets. Actually, the sketch provides a type of self-supervised signal to guide the learning of the model. Moreover, we propose to construct an auxiliary self-supervised task by classifying real/fake samples, in order to further improve the representation ability of the encoder. We thoroughly evaluate the performance of ALMS on both single-label and multi-label classification tasks, and the results demonstrate its superior performance against the state-of-the-art methods. The code can be found at https://github.com/lrq99/ALMS.

The existing neural architecture search (NAS) methods usually restrict the search space to the pre-defined types of block for a fixed macro-architecture. However, this strategy will limit the search space and affect architecture flexibility if block proposal search (BPS) is not considered for NAS. As a result, block structure search is the bottleneck in many previous NAS works. In this work, we propose a new evolutionary algorithm referred to as latency EvoNAS (LEvoNAS) for block structure search, and also incorporate it to the NAS framework by developing a novel two-stage framework referred to as Block Proposal NAS (BP-NAS). Comprehensive experimental results on two computer vision tasks demonstrate the superiority of our newly proposed approach over the state-of-the-art lightweight methods. For the classification task on the ImageNet dataset, our BPN-A is better than 1.0-MobileNetV2 with similar latency, and our BPN-B saves 23.7% latency when compared with 1.4-MobileNetV2 with higher top-1 accuracy. Furthermore, for the object detection task on the COCO dataset, our method achieves significant performance improvement than MobileNetV2, which demonstrates the generalization capability of our newly proposed framework.

There are two major lines of works, i.e., anchor-based and frame-based approaches, in the field of temporal action localization. But each line of works is inherently limited to a certain detection granularity and cannot simultaneously achieve high recall rates with accurate action boundaries. In this work, we propose a progressive cross-granularity cooperation (PCG-TAL) framework to effectively take advantage of complementarity between the anchor-based and frame-based paradigms, as well as between two-view clues (i.e., appearance and motion). Specifically, our new Anchor-Frame Cooperation (AFC) module can effectively integrate both two-granularity and two-stream knowledge at the feature and proposal levels, as well as within each AFC module and across adjacent AFC modules. Specifically, the RGB-stream AFC module and the flow-stream AFC module are stacked sequentially to form a progressive localization framework. The whole framework can be learned in an end-to-end fashion, whilst the temporal action localization performance can be gradually boosted in a progressive manner. Our newly proposed framework outperforms the state-of-the-art methods on three benchmark datasets the THUMOS14, ActivityNet v1.3 and UCF-101-24, which clearly demonstrates the effectiveness of our framework.

Weakly supervised temporal action localization is a challenging task as only the video-level annotation is available during the training process. To address this problem, we propose a two-stage approach to generate high-quality frame-level pseudo labels by fully exploiting multi-resolution information in the temporal domain and complementary information between the appearance (i.e., RGB) and motion (i.e., optical flow) streams. In the first stage, we propose an Initial Label Generation (ILG) module to generate reliable initial frame-level pseudo labels. Specifically, in this newly proposed module, we exploit temporal multi-resolution consistency and cross-stream consistency to generate high quality class activation sequences (CASs), which consist of a number of sequences with each sequence measuring how likely each video frame belongs to one specific action class. In the second stage, we propose a Progressive Temporal Label Refinement (PTLR) framework to iteratively refine the pseudo labels, in which we use a set of selected frames with highly confident pseudo labels to progressively train two networks and better predict action class scores at each frame. Specifically, in our newly proposed PTLR framework, two networks called Network-OTS and Network-RTS, which are respectively used to generate CASs for the original temporal scale and the reduced temporal scales, are used as two streams (i.e., the OTS stream and the RTS stream) to refine the pseudo labels in turn. By this way, multi-resolution information in the temporal domain is exchanged at the pseudo label level, and our work can help improve each network/stream by exploiting the refined pseudo labels from another network/stream. Comprehensive experiments on two benchmark datasets THUMOS14 and ActivityNet v1.3 demonstrate the effectiveness of our newly proposed method for weakly supervised temporal action localization.

In this work, we propose a new two-view domain adaptation network named Deep-Shallow Domain Adaptation Network (DSDAN) for 3D point cloud recognition. Different from the traditional 2D image recognition task, the valuable texture information is often absent in point cloud data, making point cloud recognition a challenging task, especially in the cross-dataset scenario where the training and testing data exhibit a considerable distribution mismatch. In our DSDAN method, we tackle the challenging cross-dataset 3D point cloud recognition task from two aspects. On one hand, we propose a two-view learning framework, such that we can effectively leverage multiple feature representations to improve the recognition performance. To this end, we propose a simple and efficient Bag-of-Points feature method, as a complementary view to the deep representation. Moreover, we also propose a cross view consistency loss to boost the two-view learning framework. On the other hand, we further propose a two-level adaptation strategy to effectively address the domain distribution mismatch issue. Specifically, we apply a feature-level distribution alignment module for each view, and also propose an instance-level adaptation approach to select highly confident pseudo-labeled target samples for adapting the model to the target domain, based on which a co-training scheme is used to integrate the learning and adaptation process on the two views. Extensive experiments on the benchmark dataset show that our newly proposed DSDAN method outperforms the existing state-of-the-art methods for the cross-dataset point cloud recognition task.

In this work, we propose a new generic multi-modality domain adaptation framework called Progressive Modality Cooperation (PMC) to transfer the knowledge learned from the source domain to the target domain by exploiting multiple modality clues (e.g., RGB and depth) under the multi-modality domain adaptation (MMDA) and the more general multi-modality domain adaptation using privileged information (MMDA-PI) settings. Under the MMDA setting, the samples in both domains have all the modalities. Through effective collaboration among multiple modalities, the two newly proposed modules in our PMC can select the reliable pseudo-labeled target samples, which captures the modality-specific information and modality-integrated information, respectively. Under the MMDA-PI setting, some modalities are missing in the target domain. Hence, to better exploit the multi-modality data in the source domain, we further propose the PMC with privileged information (PMC-PI) method by proposing a new multi-modality data generation (MMG) network. MMG generates the missing modalities in the target domain based on the source domain data by considering both domain distribution mismatch and semantics preservation, which are respectively achieved by using adversarial learning and conditioning on weighted pseudo semantic class labels. Extensive experiments on three image datasets and eight video datasets for various multi-modality cross-domain visual recognition tasks under both MMDA and MMDA-PI settings clearly demonstrate the effectiveness of our proposed PMC framework.

Recently, dense video captioning has made attractive progress in detecting and captioning all events in a long untrimmed video. Despite promising results were achieved, most existing methods do not sufficiently explore the scene evolution within an event temporal proposal for captioning, and therefore perform less satisfactorily when the scenes and objects change over a relatively long proposal. To address this problem, we propose a graph-based partition-and-summarization (GPaS) framework for dense video captioning within two stages. For the “partition” stage, a whole event proposal is split into short video segments for captioning at a finer level. For the “summarization” stage, the generated sentences carrying rich description information for each segment are summarized into one sentence to describe the whole event. We particularly focus on the “summarization” stage, and propose a framework that effectively exploits the relationship between semantic words for summarization. We achieve this goal by treating semantic words as the nodes in a graph and learning their interactions by coupling Graph Convolutional Network (GCN) and Long Short Term Memory (LSTM), with the aid of visual cues. Two schemes of GCN-LSTM Interaction (GLI) modules are proposed for seamless integration of GCN and LSTM. The effectiveness of our approach is demonstrated via an extensive comparison with the state-of-the-arts methods on the two benchmarks ActivityNet Captions dataset and YouCook II dataset.

In this work, we propose a new layer-by-layer channel pruning method called Channel Pruning guided by classification Loss and feature Importance (CPLI). In contrast to the existing layer-by-layer channel pruning approaches that only consider how to reconstruct the features from the next layer, our approach additionally take the classification loss into account in the channel pruning process. We also observe that some reconstructed features will be removed at the next pruning stage. So it is unnecessary to reconstruct these features. To this end, we propose a new strategy to suppress the influence of unimportant features (i.e., the features will be removed at the next pruning stage). Our comprehensive experiments on three benchmark datasets, i.e., CIFAR-10, ImageNet, and UCF-101, demonstrate the effectiveness of our CPLI method.

In this work, we propose a unified model compression framework called Multi-Dimensional Pruning (MDP) to simultaneously compress the convolutional neural networks (CNNs) on multiple dimensions. In contrast to the existing model compression methods that only aim to reduce the redundancy along either the spatial/spatial-temporal dimension (e.g., spatial dimension for 2D CNNs, spatial and temporal dimensions for 3D CNNs) or the channel dimension, our newly proposed approach can simultaneously reduce the spatial/spatial-temporal and the channel redundancies for CNNs. Specifically, in order to reduce the redundancy along the spatial/spatial-temporal dimension, we downsample the input tensor of a convolutional layer, in which the scaling factor for the downsampling operation is adaptively selected by our approach. After the convolution operation, the output tensor is upsampled to the original size to ensure the unchanged input size for the subsequent CNN layers. To reduce the channel-wise redundancy, we introduce a gate for each channel of the output tensor as its importance score, in which the gate value is automatically learned. The channels with small importance scores will be removed after the model compression process. Our comprehensive experiments on four benchmark datasets demonstrate that our MDP framework outperforms the existing methods when pruning both 2D CNNs and 3D CNNs.

This tutorial aims at reviewing the recent progress in the deep learning based data compression, including image compression and video compression. In the past years, deep learning techniques have been successfully applied to a large number of computer vision and image processing tasks. However, for the data compression task, the traditional approaches (i.e., block based motion estimation and motion compensation, etc.) are still widely employed in the mainstream codecs. Considering the powerful representation capability, it is possible to improve the data compression performance by employing the advanced deep learning technologies. To this end, deep leaning based compression approaches have recently received significant attention from both academia and industry in the field of computer vision and image/video compression. In this tutorial, we will introduce the related deep learning techniques for image compression and video compression. Specifically, in this tutorial, we will first introduce the basic pipeline for the traditional codecs, such as JPEG, H.264 and HEVC. Then, we will discuss the common network architectures for visual data compression and analyse different learning based entropy models. Based on these techniques, we will describe several widely used end-to-end optimized frameworks for visual data compression. In summary, our tutorial will cover both the traditional data coding techniques and the popular learning based visual data compression algorithms, which will help the audiences with different backgrounds learn the recent progresses in this emerging research area.

Subspace clustering is an effective method that has been successfully applied to many applications. Here, we propose a novel subspace clustering model for multi-view data using a latent representation termed Latent Multi-View Subspace Clustering (LMSC). Unlike most existing single-view subspace clustering methods, which directly reconstruct data points using original features, our method explores underlying complementary information from multiple views and simultaneously seeks the underlying latent representation. Using the complementarity of multiple views, the latent representation depicts data more comprehensively than each individual view, accordingly making subspace representation more accurate and robust. We proposed two LMSC formulations: linear LMSC (lLMSC), based on linear correlations between latent representation and each view, and generalized LMSC (gLMSC), based on neural networks to handle general relationships. The proposed method can be efficiently optimized under the Augmented Lagrangian Multiplier with Alternating Direction Minimization (ALM-ADM) framework. Extensive experiments on diverse datasets demonstrate the effectiveness of the proposed method.

Object localization and segmentation in weakly labeled videos are two interesting yet challenging tasks. Models built for simultaneous object localization and segmentation have been explored in the conventional fully supervised learning scenario to boost the performance of each task. However, none of the existing works has attempted to jointly learn object localization and segmentation models under weak supervision. To this end, we propose a joint learning framework called Self-Paced Fine-Tuning Network (SPFTN) for localizing and segmenting objects in weakly labelled videos. Learning the deep model jointly for object localization and segmentation under weak supervision is very challenging as the learning process of each single task would face serious ambiguity issue due to the lack of bounding-box or pixel-level supervision. To address this problem, our proposed deep SPFTN model is carefully designed with a novel multi-task self-paced learning objective, which leverages the task-specific prior knowledge and the knowledge that has been already captured to infer the confident training samples for each task. By aggregating the confident knowledge from each single task to mine reliable patterns and learning deep feature representation for both tasks, the proposed learning framework can address the ambiguity issue under weak supervision with simple optimization. Comprehensive experiments on the large-scale YouTube-Objects and DAVIS datasets demonstrate that the proposed approach achieves superior performance when compared with other state-of-the-art methods and the baseline networks/models.

Recently, the research field of salient object detection is undergoing a rapid and remarkable development along with the wide usage of deep neural networks. Being trained with a large number of images annotated with strong pixel-level ground-truth masks, the deep salient object detectors have achieved the state-of-the-art performance. However, it is expensive and time-consuming to provide the pixel-level ground-truth masks for each training image. To address this problem, this paper proposes one of the earliest frameworks to learn deep salient object detectors without requiring any human annotation. The supervisory signals used in our learning framework are generated through a novel supervision synthesis scheme, in which the key insights are "knowledge source transition" and "supervision by fusion". Specifically, in the proposed learning framework, both the external knowledge source and the internal knowledge source are explored dynamically to provide informative cues for synthesizing supervision required in our approach, while a two-stream fusion mechanism is also established to implement the supervision synthesis process. Comprehensive experiments on four benchmark datasets demonstrate that the deep salient object detector trained by our newly proposed learning framework often works well without requiring any human annotated masks, which even approaches to its upper-bound obtained under the fully supervised learning fashion (within only 3 percent performance gap). Besides, we also apply the salient object detector learnt with our annotation-free learning framework to assist the weakly supervised semantic segmentation task, which demonstrates that our approach can also alleviate the heavy supplementary supervision required in the existing weakly supervised semantic segmentation framework.

In this paper, a new metric learning algorithm is proposed to improve face verification and person re-identification in RGB images by learning from RGB and Depth (RGB-D) training images. We address this problem by formulating it as a Learning Using Privileged Information problem, in which the additional depth images associated with the RGB training images are not available for the testing process. Based on the large margin nearest neighbor (LMNN) classification framework, we propose an effective metric learning method called large margin nearest neighbor classification with privileged information (LMNN+) by incorporating depth information to improve decision function learning in the training process. Specifically, two distance metrics based on visual features as well as depth features are jointly learned by minimizing the triplet loss in which the within-class difference is minimized, while the between-class difference is maximized. The distances in the depth feature space can be utilized to guide the training process in the visual feature space. In addition, we propose an efficient optimization method which can handle billions of constraints in the optimization problem of LMNN+. The comprehensive experiments on the EUROCOM data set, the CurtinFaces data set as well as the BIWI RGBD-ID data set demonstrate the effectiveness of our algorithm for face verification and person re-identification by leveraging privileged information.

In this work, we propose a division-and-summarization (DaS) framework for dense video captioning. After partitioning each untrimmed long video as multiple event proposals, where each event proposal consists of a set of short video segments, we extract visual feature (e.g., C3D feature) from each segment and use the existing image/video captioning approach to generate one sentence description for this segment. Considering that the generated sentences contain rich semantic descriptions about the whole event proposal, we formulate the dense video captioning task as a visual cue aided sentence summarization problem and propose a new two stage Long Short Term Memory (LSTM) approach equipped with a new hierarchical attention mechanism to summarize all generated sentences as one descriptive sentence with the aid of visual features. Specifically, the first-stage LSTM network takes all semantic words from the generated sentences and the visual features from all segments within one event proposal as the input, and acts as the encoder to effectively summarize both semantic and visual information related to this event proposal. The second-stage LSTM network takes the output from the first-stage LSTM network and the visual features from all video segments within one event proposal as the input, and acts as the decoder to generate one descriptive sentence for this event proposal. Our comprehensive experiments on the ActivityNet Captions dataset demonstrate the effectiveness of our newly proposed DaS framework for dense video captioning.

Video compression algorithms are widely used to reduce the huge size of video data, but they also introduce unpleasant visual artifacts due to the lossy compression. In order to improve the quality of the compressed videos, we proposed a deep non-local Kalman network for compression artifact reduction. Specifically, the video restoration is modeled as a Kalman filtering procedure and the decoded frames can be restored from the proposed deep Kalman model. Instead of using the noisy previous decoded frames as temporal information, the less noisy previous restored frame is employed in a recursive way, which provides the potential to generate high quality restored frames. In the proposed framework, several deep neural networks are utilized to estimate the corresponding states in the Kalman filter and integrated together in the deep Kalman filtering network. More importantly, we also exploit the non-local prior information by incorporating the spatial and temporal non-local networks for better restoration. Our approach takes the advantages of both the model-based methods and learning-based methods, by combining the recursive nature of the Kalman model and powerful representation ability of neural networks. Extensive experimental results on the Vimeo-90k and HEVC benchmark datasets demonstrate the effectiveness of our proposed method.

No abstract available.

Conventional video compression approaches use the predictive coding architecture and encode the corresponding motion information and residual information. In this paper, taking advantage of both classical architecture in the conventional video compression method and the powerful non-linear representation ability of neural networks, we propose the first end-to-end video compression deep model that jointly optimizes all the components for video compression. Specifically, learning based optical flow estimation is utilized to obtain the motion information and reconstruct the current frames. Then we employ two auto-encoder style neural networks to compress the corresponding motion and residual information. All the modules are jointly learned through a single loss function, in which they collaborate with each other by considering the trade-off between reducing the number of compression bits and improving quality of the decoded video. Experimental results show that the proposed approach can outperform the widely used video coding standard H.264 in terms of PSNR and be even on par with the latest standard H.265 in terms of MS-SSIM. Code is released at https://github.com/GuoLusjtu/DVC.

Spatio-temporal action localization consists of three levels of tasks: spatial localization, action classification, and temporal segmentation. In this work, we propose a new Progressive Cross-stream Cooperation (PCSC) framework to iterative improve action localization results and generate better bounding boxes for one stream (i.e., Flow/RGB) by leveraging both region proposals and features from another stream (i.e., RGB/Flow) in an iterative fashion. Specifically, we first generate a larger set of region proposals by combining the latest region proposals from both streams, from which we can readily obtain a larger set of labelled training samples to help learn better action detection models. Second, we also propose a new message passing approach to pass information from one stream to another stream in order to learn better representations, which also leads to better action detection models. As a result, our iterative framework progressively improves action localization results at the frame level. To improve action localization results at the video level, we additionally propose a new strategy to train class-specific actionness detectors for better temporal segmentation, which can be readily learnt by using the training samples around temporal boundaries. Comprehensive experiments on two benchmark datasets UCF-101-24 and J-HMDB demonstrate the effectiveness of our newly proposed approaches for spatio-temporal action localization in realistic scenarios.

Generative adversarial networks (GANs) have demonstrated great success in generating various visual content. However, images generated by existing GANs are often of attributes (e.g., smiling expression) learned from one image domain. As a result, generating images of multiple attributes requires many real samples possessing multiple attributes which are very resource expensive to be collected. In this paper, we propose a novel GAN, namely IntersectGAN, to learn multiple attributes from different image domains through an intersecting architecture. For example, given two image domains $X_1$ and $X_2$ with certain attributes, the intersection $X_1 \cap X_2$ denotes a new domain where images possess the attributes from both $X_1$ and $X_2$ domains. The proposed IntersectGAN consists of two discriminators $D_1$ and $D_2$ to distinguish between generated and real samples of different domains, and three generators where the intersection generator is trained against both discriminators. And an overall adversarial loss function is defined over three generators. As a result, our proposed IntersectGAN can be trained on multiple domains of which each presents one specific attribute, and eventually eliminates the need of real sample images simultaneously possessing multiple attributes. By using the CelebFaces Attributes dataset, our proposed IntersectGAN is able to produce high quality face images possessing multiple attributes (e.g., a face with black hair and a smiling expression). Both qualitative and quantitative evaluations are conducted to compare our proposed IntersectGAN with other baseline methods. Besides, several different applications of IntersectGAN have been explored with promising results.

In this paper, we address the problem of large-scale graph-based semi-supervised learning for multi-class classification. Most existing scalable graph-based semi-supervised learning methods are based on the hard linear constraint or cannot cope with the unseen samples, which limits their applications and learning performance. To this end, we build upon our previous work flexible manifold embedding (FME) [1] and propose two novel linear-complexity algorithms called fast flexible manifold embedding (f-FME) and reduced flexible manifold embedding (r-FME). Both of the proposed methods accelerate FME and inherit its advantages. Specifically, our methods address the hard linear constraint problem by combining a regression residue term and a manifold smoothness term jointly, which naturally provides the prediction model for handling unseen samples. To reduce computational costs, we exploit the underlying relationship between a small number of anchor points and all data points to construct the graph adjacency matrix, which leads to simplified closed-form solutions. The resultant f-FME and r-FME algorithms not only scale linearly in both time and space with respect to the number of training samples but also can effectively utilize information from both labeled and unlabeled data. Experimental results show the effectiveness and scalability of the proposed methods.

The performance of object detection has recently been significantly improved due to the powerful features learnt through convolutional neural networks (CNNs). Despite the remarkable success, there are still several major challenges in object detection, including object rotation, within-class diversity, and between-class similarity, which generally degenerate object detection performance. To address these issues, we build up the existing state-of-the-art object detection systems and propose a simple but effective method to train rotation-invariant and Fisher discriminative CNN models to further boost object detection performance. This is achieved by optimizing a new objective function that explicitly imposes a rotation-invariant regularizer and a Fisher discrimination regularizer on the CNN features. Specifically, the first regularizer enforces the CNN feature representations of the training samples before and after rotation to be mapped closely to each other in order to achieve rotation-invariance. The second regularizer constrains the CNN features to have small within-class scatter but large between-class separation. We implement our proposed method under four popular object detection frameworks, including region-CNN (R-CNN), Fast R- CNN, Faster R- CNN, and R- FCN. In the experiments, we comprehensively evaluate the proposed method on the PASCAL VOC 2007 and 2012 data sets and a publicly available aerial image data set. Our proposed methods outperform the existing baseline methods and achieve the state-of-the-art results.

Most recent learning-based single-image superresolution methods first interpolate the low-resolution (LR) input, from which overlapped LR features are then extracted to reconstruct their high-resolution (HR) counterparts and the final HR image. However, most of them neglect to take advantage of the intermediate recovered HR image to enhance image quality further. We conduct principal component analysis (PCA) to reduce LR feature dimension. Then we find that the number of principal components after conducting PCA in the LR feature space from the reconstructed images is larger than that from the interpolated images by using bicubic interpolation. Based on this observation, we present an unsophisticated yet effective framework named collaborative representation cascade (CRC) that learns multilayer mapping models between LR and HR feature pairs. In particular, we extract the features from the intermediate recovered image to upscale and enhance LR input progressively. In the learning phase, for each cascade layer, we use the intermediate recovered results and their original HR counterparts to learn single-layer mapping model. Then, we use this single-layer mapping model to super-resolve the original LR inputs. And the intermediate HR outputs are regarded as training inputs for the next cascade layer, until we obtain multilayer mapping models. In the reconstruction phase, we extract multiple sets of LR features from the LR image and intermediate recovered. Then, in each cascade layer, mapping model is utilized to pursue HR image. Our experiments on several commonly used image SR testing datasets show that our proposed CRC method achieves state-of-the-art image SR results.

In this paper, we propose a new unsupervised domain adaptation approach called Collaborative and Adversarial Network (CAN) through domain-collaborative and domain-adversarial training of neural networks. We add several domain classifiers on multiple CNN feature extraction blocks1, in which each domain classifier is connected to the hidden representations from one block and one loss function is defined based on the hidden presentation and the domain labels (e.g., source and target). We design a new loss function by integrating the losses from all blocks in order to learn domain informative representations from lower blocks through collaborative learning and learn domain uninformative representations from higher blocks through adversarial learning. We further extend our CAN method as Incremental CAN (iCAN), in which we iteratively select a set of pseudo-labelled target samples based on the image classifier and the last domain classifier from the previous training epoch and re-train our CAN model by using the enlarged training set. Comprehensive experiments on two benchmark datasets Office and ImageCLEF-DA clearly demonstrate the effectiveness of our newly proposed approaches CAN and iCAN for unsupervised domain adaptation.

Cross-scene hyperspectral image (HSI) classification has recently become increasingly popular due to its crucial use in various applications. It poses great challenges to existing domain adaptation methods because of the data set shift, that is, two scenes exhibit huge distribution discrepancy. To tackle this problem, we propose a new domain adaptation method called hyperspectral feature adaptation and augmentation (HFAA) for cross-scene HSI classification. The proposed HFAA method learns a common subspace by introducing two different projection matrices to extract the transferable knowledge from the source domain to the target domain. To further enhance the common subspace representation, we propose to augment it by the feature selection strategy. HFAA can make full use of the original features from both source and target domains, and increase the similarity of the samples with the same label from the two domains. Our proposed HFAA method achieves compact but discriminative feature representations, which make it well suited for data sets with a large number of classes and huge interclass ambiguity. Experimental results on the Earth Observing 1 hyperspectral data set show that HFAA can produce state-of-the-art performance and surpass previous methods.

In this work, we propose a framework for recognizing RGB images or videos by learning from RGB-D training data that contains additional depth information. We formulate this task as a new unsupervised domain adaptation (UDA) problem, in which we aim to take advantage of the additional depth features in the source domain and also cope with the data distribution mismatch between the source and target domains. To handle the domain distribution mismatch, we propose to learn an optimal projection matrix to map the samples from both domains into a common subspace such that the domain distribution mismatch can be reduced. Such projection matrix can be effectively optimized by exploiting different strategies. Moreover, we also use different ways to utilize the additional depth features. To simultaneously cope with the above two issues, we formulate a unified learning framework called domain adaptation from multi-view to single-view (DAM2S). By defining various forms of regularizers in our DAM2S framework, different strategies can be readily incorporated to learn robust SVM classifiers for classifying the target samples, and three methods are developed under our DAM2S framework. We conduct comprehensive experiments for object recognition, cross-dataset and cross-view action recognition, which demonstrate the effectiveness of our proposed methods for recognizing RGB images and videos by learning from RGB-D data.

Domain adaptation between diverse source and target domains is challenging, especially in the real-world visual recognition tasks where the images and videos consist of significant variations in viewpoints, illuminations, qualities, etc. In this paper, we propose a new approach for domain generalization and domain adaptation based on exemplar SVMs. Specifically, we decompose the source domain into many subdomains, each of which contains only one positive training sample and all negative samples. Each subdomain is relatively less diverse, and is expected to have a simpler distribution. By training one exemplar SVM for each subdomain, we obtain a set of exemplar SVMs. To further exploit the inherent structure of source domain, we introduce a nuclear-norm based regularizer into the objective function in order to enforce the exemplar SVMs to produce a low-rank output on training samples. In the prediction process, the confident exemplar SVM classifiers are selected and reweigted according to the distribution mismatch between each subdomain and the test sample in the target domain. We formulate our approach based on the logistic regression and least square SVM algorithms, which are referred to as low rank exemplar SVMs (LRE-SVMs) and low rank exemplar least square SVMs (LRE-LSSVMs), respectively. A fast algorithm is also developed for accelerating the training of LRE-LSSVMs. We further extend Domain Adaptation Machine (DAM) to learn an optimal target classifier for domain adaptation, and show that our approach can also be applied to domain adaptation with evolving target domain, where the target data distribution is gradually changing. The comprehensive experiments for object recognition and action recognition demonstrate the effectiveness of our approach for domain generalization and domain adaptation with fixed and evolving target domains.

Skeleton-based human action recognition has attracted a lot of research attention during the past few years. Recent works attempted to utilize recurrent neural networks to model the temporal dependencies between the 3D positional configurations of human body joints for better analysis of human activities in the skeletal data. The proposed work extends this idea to spatial domain as well as temporal domain to better analyze the hidden sources of action-related information within the human skeleton sequences in both of these domains simultaneously. Based on the pictorial structure of Kinect's skeletal data, an effective tree-structure based traversal framework is also proposed. In order to deal with the noise in the skeletal data, a new gating mechanism within LSTM module is introduced, with which the network can learn the reliability of the sequential data and accordingly adjust the effect of the input data on the updating procedure of the long-term context representation stored in the unit's memory cell. Moreover, we introduce a novel multi-modal feature fusion strategy within the LSTM unit in this paper. The comprehensive experimental results on seven challenging benchmark datasets for human action recognition demonstrate the effectiveness of the proposed method.

GoDec is an efficient low-rank matrix decomposition algorithm. However, optimal performance depends on sparse errors and Gaussian noise. This paper aims to address the problem that a matrix is composed of a low-rank component and unknown corruptions. We introduce a robust local similarity measure called correntropy to describe the corruptions and, in doing so, obtain a more robust and faster low-rank decomposition algorithm: GoDec+. Based on half-quadratic optimization and greedy bilateral paradigm, we deliver a solution to the maximum correntropy criterion (MCC)-based low-rank decomposition problem. Experimental results show that GoDec+ is efficient and robust to different corruptions including Gaussian noise, Laplacian noise, salt & pepper noise, and occlusion on both synthetic and real vision data. We further apply GoDec+ to more general applications including classification and subspace clustering. For classification, we construct an ensemble subspace from the low-rank GoDec+ matrix and introduce an MCC-based classifier. For subspace clustering, we utilize GoDec+ values low-rank matrix for MCC-based self-expression and combine it with spectral clustering. Face recognition, motion segmentation, and face clustering experiments show that the proposed methods are effective and robust. In particular, we achieve the state-of-the-art performance on the Hopkins 155 data set and the first 10 subjects of extended Yale B for subspace clustering.

In this paper, we propose a new exemplar-based multi-view domain generalization (EMVDG) framework for visual recognition by learning robust classifier that are able to generalize well to arbitrary target domain based on the training samples with multiple types of features (i.e., multi-view features). In this framework, we aim to address two issues simultaneously. First, the distribution of training samples (i.e., the source domain) is often considerably different from that of testing samples (i.e., the target domain), so the performance of the classifiers learnt on the source domain may drop significantly on the target domain. Moreover, the testing data are often unseen during the training procedure. Second, when the training data are associated with multi-view features, the recognition performance can be further improved by exploiting the relation among multiple types of features. To address the first issue, considering that it has been shown that fusing multiple SVM classifiers can enhance the domain generalization ability, we build our EMVDG framework upon exemplar SVMs (ESVMs), in which a set of ESVM classifiers are learnt with each one trained based on one positive training sample and all the negative training samples. When the source domain contains multiple latent domains, the learnt ESVM classifiers are expected to be grouped into multiple clusters. To address the second issue, we propose two approaches under the EMVDG framework based on the consensus principle and the complementary principle, respectively. Specifically, we propose an EMVDG_CO method by adding a co-regularizer to enforce the cluster structures of ESVM classifiers on different views to be consistent based on the consensus principle. Inspired by multiple kernel learning, we also propose another EMVDG_MK method by fusing the ESVM classifiers from different views based on the complementary principle. In addition, we further extend our EMVDG framework to exemplar-based multi-view domain adaptation (EMVDA) framework when the unlabeled target domain data are available during the training procedure. The effectiveness of our EMVDG and EMVDA frameworks for visual recognition is clearly demonstrated by comprehensive experiments on three benchmark data sets.

Object segmentation in weakly labelled videos is an interesting yet challenging task, which aims at learning to perform category-specific video object segmentation by only using video-level tags. Existing works in this research area might still have some limitations, e.g., lack of effective DNN-based learning frameworks, under-exploring the context information, and requiring to leverage the unstable negative video collection, which prevent them from obtaining more promising performance. To this end, we propose a novel self-paced fine-tuning network (SPFTN)-based framework, which could learn to explore the context information within the video frames and capture adequate object semantics without using the negative videos. To perform weakly supervised learning based on the deep neural network, we make the earliest effort to integrate the self-paced learning regime and the deep neural network into a unified and compatible framework, leading to the self-paced fine-tuning network. Comprehensive experiments on the large-scale YouTube-Objects and DAVIS datasets demonstrate that the proposed approach achieves superior performance as compared with other state-of-the-art methods as well as the baseline networks and models.

The goal of complex event detection is to automatically detect whether an event of interest happens in temporally untrimmed long videos which usually consist of multiple video shots. Observing some video shots in positive (resp. negative) videos are irrelevant (resp. relevant) to the given event class, we formulate this task as a multi-instance learning (MIL) problem by taking each video as a bag and the video shots in each video as instances. To this end, we propose a new MIL method, which simultaneously learns a linear SVM classifier and infers a binary indicator for each instance in order to select reliable training instances from each positive or negative bag. In our new objective function, we balance the weighted training errors and a l1-l2mixed-norm regularization term which adaptively selects reliable shots as training instances from different videos to have them as diverse as possible. We also develop an alternating optimization approach that can efficiently solve our proposed objective function. Extensive experiments on the challenging real-world Multimedia Event Detection (MED) datasets MEDTest-14, MEDTest-13 and CCV clearly demonstrate the effectiveness of our proposed MIL approach for complex event detection.

Visual emotion recognition aims to associate images with appropriate emotions. There are different visual stimuli that can affect human emotion from low-level to high-level, such as color, texture, part, object, etc. However, most existing methods treat different levels of features as independent entity without having effective method for feature fusion. In this paper, we propose a unified CNN-RNN model to predict the emotion based on the fused features from different levels by exploiting the dependency among them. Our proposed architecture leverages convolutional neural network (CNN) with multiple layers to extract different levels of features with in a multi-task learning framework, in which two related loss functions are introduced to learn the feature representation. Considering the dependencies within the low-level and high-level features, a new bidirectional recurrent neural network (RNN) is proposed to integrate the learned features from different layers in the CNN model. Extensive experiments on both Internet images and art photo datasets demonstrate that our method outperforms the state-of-the-art methods with at least 7% performance improvement.

We present an RGB and Depth (RGBD) video segmentation method that takes advantage of depth data and can extract multiple foregrounds in the scene. This video segmentation is addressed as an object proposal selection problem formulated in a fully-connected graph, where a flexible number of foregrounds may be chosen. In our graph, each node represents a proposal, and the edges model intra-frame and inter-frame constraints on the solution. The proposals are selected based on an RGBD video saliency map in which depth-based features are utilized to enhance the identification of foregrounds. Experiments show that the proposed multiple foreground segmentation method outperforms related techniques, and the depth cue serves as a helpful complement to RGB features. Moreover, our method provides performance comparable to the state-of-the-art RGB video segmentation techniques on regular RGB videos with estimated depth maps.

In this paper, a weakly supervised domain generalization (WSDG) method is proposed for real-world visual recognition tasks, in which we train classifiers by using Web data (e.g., Web images and Web videos) with noisy labels. In particular, two challenging problems need to be solved when learning robust classifiers, in which the first issue is to cope with the label noise of training Web data from the source domain, while the second issue is to enhance the generalization capability of learned classifiers to an arbitrary target domain. In order to handle the first problem, the training samples within each category are partitioned into clusters, where we use one bag to denote each cluster and instances to denote the samples in each cluster. Then, we identify a proportion of good training samples in each bag and train robust classifiers by using the good training samples, which leads to a multi-instance learning (MIL) problem. In order to handle the second problem, we assume that the training samples possibly form a set of hidden domains, with each hidden domain associated with a distinctive data distribution. Then, for each category and each hidden latent domain, we propose to learn one classifier by extending our MIL formulation, which leads to our WSDG approach. In the testing stage, our approach can obtain better generalization capability by effectively integrating multiple classifiers from different latent domains in each category. Moreover, our WSDG approach is further extended to utilize additional textual descriptions associated with Web data as privileged information (PI), although testing data do not have such PI. Extensive experiments on three benchmark data sets indicate that our newly proposed methods are effective for real-world visual recognition tasks by learning from Web data.

In this paper, we propose new approaches for action and event recognition by leveraging a large number of freely available Web videos (e.g., from Flickr video search engine) and Web images (e.g., from Bing and Google image search engines). We address this problem by formulating it as a new multi-domain adaptation problem, in which heterogeneous Web sources are provided. Specifically, we are given different types of visual features (e.g., the DeCAF features from Bing/Google images and the trajectory-based features from Flickr videos) from heterogeneous source domains and all types of visual features from the target domain. Considering the target domain is more relevant to some source domains, we propose a new approach named multi-domain adaptation with heterogeneous sources (MDA-HS) to effectively make use of the heterogeneous sources. In MDA-HS, we simultaneously seek for the optimal weights of multiple source domains, infer the labels of target domain samples, and learn an optimal target classifier. Moreover, as textual descriptions are often available for both Web videos and images, we propose a novel approach called MDA-HS using privileged information (MDA-HS+) to effectively incorporate the valuable textual information into our MDA-HS method, based on the recent learning using privileged information paradigm. MDA-HS+ can be further extended by using a new elastic-net-like regularization. We solve our MDA-HS and MDA-HS+ methods by using the cutting-plane algorithm, in which a multiple kernel learning problem is derived and solved. Extensive experiments on three benchmark data sets demonstrate that our proposed approaches are effective for action and event recognition without requiring any labeled samples from the target domain.

The SVM+ approach has shown excellent performance in visual recognition tasks for exploiting privileged information in the training data. In this paper, we propose two efficient algorithms for solving the linear and kernel SVM+, respectively. For linear SVM+, we absorb the bias term into the weight vector, and formulate a new optimization problem with simpler constraints in the dual form. Then, we develop an efficient dual coordinate descent algorithm to solve the new optimization problem. For kernel SVM+, we further apply the l2-loss, which leads to a simpler optimization problem in the dual form with only half of dual variables when compared with the dual form of the original SVM+ method. More interestingly, we show that our new dual problem can be efficiently solved by using the SMO algorithm of the one-class SVM problem. Comprehensive experiments on three datasets clearly demonstrate that our proposed algorithms achieve significant speed-up than the state-of-the-art solvers for linear and kernel SVM+.

Trace-norm regularization plays an important role in many areas such as computer vision and machine learning. When solving general large-scale trace-norm regularized problems, existing methods may be computationally expensive due to many high-dimensional truncated singular value decompositions (SVDs) or the unawareness of matrix ranks. In this paper, we propose a proximal Riemannian pursuit (PRP) paradigm which addresses a sequence of trace-norm regularized subproblems defined on nonlinear matrix varieties. To address the subproblem, we extend the proximal gradient method on vector space to nonlinear matrix varieties, in which the SVDs of intermediate solutions are maintained by cheap low-rank QR decompositions, therefore making the proposed method more scalable. Empirical studies on several tasks, such as matrix completion and low-rank representation based subspace clustering, demonstrate the competitive performance of the proposed paradigms over existing methods.

The k-dimensional coding schemes refer to a collection of methods that attempt to represent data using a set of representative k-dimensional vectors and include nonnegative matrix factorization, dictionary learning, sparse coding, k-means clustering, and vector quantization as special cases. Previous generalization bounds for the reconstruction error of the k-dimensional coding schemes are mainly dimensionality-independent. A major advantage of these bounds is that they can be used to analyze the generalization error when data are mapped into an infinite- or high-dimensional feature space. However, many applications use finite-dimensional data features. Can we obtain dimensionality-dependent generalization bounds for k-dimensional coding schemes that are tighter than dimensionality-independent bounds when data are in a finite-dimensional feature space? Yes. In this letter, we address this problem and derive a dimensionality-dependent generalization bound for k-dimensional coding schemes by bounding the covering number of the loss function class induced by the reconstruction error. The bound is of order [Formula: see text], where m is the dimension of features, k is the number of the columns in the linear implementation of coding schemes, and n is the size of sample, [Formula: see text] when n is finite and [Formula: see text] when n is infinite. We show that our bound can be tighter than previous results because it avoids inducing the worst-case upper bound on k of the loss function. The proposed generalization bound is also applied to some specific coding schemes to demonstrate that the dimensionality-dependent bound is an indispensable complement to the dimensionality-independent generalization bounds.

It is often expensive and time consuming to collect labeled training samples in many real-world applications. To reduce human effort on annotating training samples, many machine learning techniques (e.g., semi-supervised learning (SSL), multi-instance learning (MIL), etc.) have been studied to exploit weakly labeled training samples. Meanwhile, when the training data is represented with multiple types of features, many multi-view learning methods have shown that classifiers trained on different views can help each other to better utilize the unlabeled training samples for the SSL task. In this paper, we study a new learning problem called multi-view weakly labeled learning, in which we aim to develop a unified approach to learn robust classifiers by effectively utilizing different types of weakly labeled multi-view data from a broad range of tasks including SSL, MIL and relative outlier detection (ROD). We propose an effective approach called co-labeling to solve the multi-view weakly labeled learning problem. Specifically, we model the learning problem on each view as a weakly labeled learning problem, which aims to learn an optimal classifier from a set of pseudo-label vectors generated by using the classifiers trained from other views. Unlike traditional co-training approaches using a single pseudo-label vector for training each classifier, our co-labeling approach explores different strategies to utilize the predictions from different views, biases and iterations for generating the pseudo-label vectors, making our approach more robust for real-world applications. Moreover, to further improve the weakly labeled learning on each view, we also exploit the inherent group structure in the pseudo-label vectors generated from different strategies, which leads to a new multi-layer multiple kernel learning problem. Promising results for text-based image retrieval on the NUS-WIDE dataset as well as news classification and text categorization on several real-world multi-view datasets clearly demonstrate that our proposed co-labeling approach achieves state-of-the-art performance for various multi-view weakly labeled learning problems including multi-view SSL, multi-view MIL and multi-view ROD.

Presents a listing of the new Associate Editors for this issue of the publication.

In this paper, we propose a novel cross-media active learning algorithm to reduce the effort on labeling images for training. The Internet images are often associated with rich textual descriptions. Even though such textual information is not available in test images, it is still useful for learning robust classifiers. In light of this, we apply the recently proposed supervised learning paradigm, learning using privileged information, to the active learning task. Specifically, we train classifiers on both visual features and privileged information, and measure the uncertainty of unlabeled data by exploiting the learned classifiers and slacking function. Then, we propose to select unlabeled samples by jointly measuring the cross-media uncertainty and the visual diversity. Our method automatically learns the optimal tradeoff parameter between the two measurements, which in turn makes our algorithms particularly suitable for real-world applications. Extensive experiments demonstrate the effectiveness of our approach.

Vision and multimedia communities have long attempted to enable computers to understand image or video content in a manner analogous to humans. Humans’ comprehension to an image or a video clip often depends on the objects that draw their attention. As a result, one fundamental and open problem is to automatically infer the attention attracting or interesting areas in an image or a video sequence. Recently, a large number of researchers explore visual saliency models to address this problem. The study on visual saliency models is originally motivated by simulating humans’ bottom-up visual attention and it is mainly based on the biological evidence that humans’ visual attention is automatically attracted by highly salient features in the visual scene, which are discriminative with respect to the surrounding environment.

Recently, low-rank representation (LRR) has shown promising performance in many real-world applications such as face clustering. However, LRR may not achieve satisfactory results when dealing with the data from nonlinear subspaces, since it is originally designed to handle the data from linear subspaces in the input space. Meanwhile, the kernel-based methods deal with the nonlinear data by mapping it from the original input space to a new feature space through a kernel-induced mapping. To effectively cope with the nonlinear data, we first propose the kernelized version of LRR in the clean data case. We also present a closed-form solution for the resultant optimization problem. Moreover, to handle corrupted data, we propose the robust kernel LRR (RKLRR) approach, and develop an efficient optimization algorithm to solve it based on the alternating direction method. In particular, we show that both the subproblems in our optimization algorithm can be efficiently and exactly solved, and it is guaranteed to obtain a globally optimal solution. Besides, our proposed algorithm can also solve the original LRR problem, which is a special case of our RKLRR when using the linear kernel. In addition, based on our new optimization technique, the kernelization of some variants of LRR can be similarly achieved. Comprehensive experiments on synthetic data sets and real-world data sets clearly demonstrate the efficiency of our algorithm, as well as the effectiveness of RKLRR and the kernelization of two variants of LRR.

We present an object-based co-segmentation method that takes advantage of depth data and is able to correctly handle noisy images in which the common foreground object is missing. With RGBD images, our method utilizes the depth channel to enhance identification of similar foreground objects via a proposed RGBD co-saliency map, as well as to improve detection of object-like regions and provide depth-based local features for region comparison. To accurately deal with noisy images where the common object appears more than or less than once, we formulate co-segmentation in a fully-connected graph structure together with mutual exclusion (mutex) constraints that prevent improper solutions. Experiments show that this object-based RGBD co-segmentation with mutex constraints outperforms related techniques on an RGBD co-segmentation dataset, while effectively processing noisy images. Moreover, we show that this method also provides performance comparable to state-of-the-art RGB co-segmentation techniques on regular RGB images with depth maps estimated from them.

In this work, we formulate a new weakly supervised domain generalization approach for visual recognition by using loosely labeled web images/videos as training data. Specifically, we aim to address two challenging issues when learning robust classifiers: 1) coping with noise in the labels of training web images/videos in the source domain; and 2) enhancing generalization capability of learnt classifiers to any unseen target domain. To address the first issue, we partition the training samples in each class into multiple clusters. By treating each cluster as a “bag” and the samples in each cluster as “instances”, we formulate a multi-instance learning (MIL) problem by selecting a subset of training samples from each training bag and simultaneously learning the optimal classifiers based on the selected samples. To address the second issue, we assume the training web images/videos may come from multiple hidden domains with different data distributions. We then extend our MIL formulation to learn one classifier for each class and each latent domain such that multiple classifiers from each class can be effectively integrated to achieve better generalization capability. Extensive experiments on three benchmark datasets demonstrate the effectiveness of our new approach for visual recognition by learning from web data.

Low rank representation (LRR) has shown promising performance for various computer vision applications such as face clustering. Existing algorithms for solving LRR usually depend on its two-variable formulation which contains the original data matrix. In this paper, we develop a fast LRR solver called FaLRR, by reformulating LRR as a new optimization problem with regard to factorized data (which is obtained by skinny SVD of the original data matrix). The new formulation benefits the corresponding optimization and theoretical analysis. Specifically, to solve the resultant optimization problem, we propose a new algorithm which is not only efficient but also theoretically guaranteed to obtain a globally optimal solution. Regarding the theoretical analysis, the new formulation is helpful for deriving some interesting properties of LRR. Last but not least, the proposed algorithm can be readily incorporated into an existing distributed framework of LRR for further acceleration. Extensive experiments on synthetic and real-world datasets demonstrate that our FaLRR achieves order-of-magnitude speedup over existing LRR solvers, and the efficiency can be further improved by incorporating our algorithm into the distributed framework of LRR.

In this paper, we propose a new multi-view domain generalization (MVDG) approach for visual recognition, in which we aim to use the source domain samples with multiple types of features (i.e., multi-view features) to learn robust classifiers that can generalize well to any unseen target domain. Considering the recent works show the domain generalization capability can be enhanced by fusing multiple SVM classifiers, we build upon exemplar SVMs to learn a set of SVM classifiers by using one positive sample and all negative samples in the source domain each time. When the source domain samples come from multiple latent domains, we expect the weight vectors of exemplar SVM classifiers can be organized into multiple hidden clusters. To exploit such cluster structure, we organize the weight vectors learnt on each view as a weight matrix and seek the low-rank representation by reconstructing this weight matrix using itself as the dictionary. To enforce the consistency of inherent cluster structures discovered from the weight matrices learnt on different views, we introduce a new regularizer to minimize the mismatch between any two representation matrices on different views. We also develop an efficient alternating optimization algorithm and further extend our MVDG approach for domain adaptation by exploiting the manifold structure of unlabeled target domain samples. Comprehensive experiments for visual recognition clearly demonstrate the effectiveness of our approaches for domain generalization and domain adaptation.

We present a framework for image classification that extends beyond the window sampling of fixed spatial pyramids and is supported by a new learning algorithm. Based on the observation that fixed spatial pyramids sample a rather limited subset of the possible image windows, we propose a method that accounts for a comprehensive set of windows densely sampled over location, size, and aspect ratio. A concise high-level image feature is derived to effectively deal with this large set of windows, and this higher level of abstraction offers both efficient handling of the dense samples and reduced sensitivity to misalignment. In addition to dense window sampling, we introduce generalized adaptive l(p)-norm multiple kernel learning (GA-MKL) to learn a robust classifier based on multiple base kernels constructed from the new image features and multiple sets of prelearned classifiers from other classes. With GA-MKL, multiple levels of image features are effectively fused, and information is shared among different classifiers. Extensive evaluation on benchmark datasets for object recognition (Caltech256 and Caltech101) and scene recognition (15Scenes) demonstrate that the proposed method outperforms the state-of-the-art under a broad range of settings.

We present a technique for multiple foreground video co-segmentation in a set of videos. This technique is based on category-independent object proposals. To identify the foreground objects in each frame, we examine the properties of the various regions that reflect the characteristics of foregrounds, considering the intra-video coherence of the foreground as well as the foreground consistency among the different videos in the set. Multiple foregrounds are handled via a multi-state selection graph in which a node representing a video frame can take multiple labels that correspond to different objects. In addition, our method incorporates an indicator matrix that for the first time allows accurate handling of cases with common foreground objects missing in some videos, thus preventing irrelevant regions from being misclassified as foreground objects. An iterative procedure is proposed to optimize our new objective function. As demonstrated through comprehensive experiments, this object-based multiple foreground video co-segmentation method compares well with related techniques that co-segment multiple foregrounds.

Given a collection of images, where each image contains several faces and is associated with a few names in the corresponding caption, the goal of face naming is to infer the correct name for each face. In this paper, we propose two new methods to effectively solve this problem by learning two discriminative affinity matrices from these weakly labeled images. We first propose a new method called regularized low-rank representation by effectively utilizing weakly supervised information to learn a low-rank reconstruction coefficient matrix while exploring multiple subspace structures of the data. Specifically, by introducing a specially designed regularizer to the low-rank representation method, we penalize the corresponding reconstruction coefficients related to the situations where a face is reconstructed by using face images from other subjects or by using itself. With the inferred reconstruction coefficient matrix, a discriminative affinity matrix can be obtained. Moreover, we also develop a new distance metric learning method called ambiguously supervised structural metric learning by using weakly supervised information to seek a discriminative distance metric. Hence, another discriminative affinity matrix can be obtained using the similarity matrix (i.e., the kernel matrix) based on the Mahalanobis distances of the data. Observing that these two affinity matrices contain complementary information, we further combine them to obtain a fused affinity matrix, based on which we develop a new iterative scheme to infer the name of each face. Comprehensive experiments demonstrate the effectiveness of our approach.

In this paper, we propose a new approach to improve face verification and person re-identification in the RGB images by leveraging a set of RGB-D data, in which we have additional depth images in the training data captured using depth cameras such as Kinect. In particular, we extract visual features and depth features from the RGB images and depth images, respectively. As the depth features are available only in the training data, we treat the depth features as privileged information, and we formulate this task as a distance metric learning with privileged information problem. Unlike the traditional face verification and person re-identification tasks that only use visual features, we further employ the extra depth features in the training data to improve the learning of distance metric in the training process. Based on the information-theoretic metric learning (ITML) method, we propose a new formulation called ITML with privileged information (ITML+) for this task. We also present an efficient algorithm based on the cyclic projection method for solving the proposed ITML+ formulation. Extensive experiments on the challenging faces data sets EUROCOM and CurtinFaces for face verification as well as the BIWI RGBD-ID data set for person re-identification demonstrate the effectiveness of our proposed approach.

In this work, we propose a new framework for recognizing RGB images captured by the conventional cameras by leveraging a set of labeled RGB-D data, in which the depth features can be additionally extracted from the depth images. We formulate this task as a new unsupervised domain adaptation (UDA) problem, in which we aim to take advantage of the additional depth features in the source domain and also cope with the data distribution mismatch between the source and target domains. To effectively utilize the additional depth features, we seek two optimal projection matrices to map the samples from both domains into a common space by preserving as much as possible the correlations between the visual features and depth features. To effectively employ the training samples from the source domain for learning the target classifier, we reduce the data distribution mismatch by minimizing the Maximum Mean Discrepancy (MMD) criterion, which compares the data distributions for each type of feature in the common space. Based on the above two motivations, we propose a new SVM based objective function to simultaneously learn the two projection matrices and the optimal target classifier in order to well separate the source samples from different classes when using each type of feature in the common space. An efficient alternating optimization algorithm is developed to solve our new objective function. Comprehensive experiments for object recognition and gender recognition demonstrate the effectiveness of our proposed approach for recognizing RGB images by learning from RGB-D data.

We present a video co-segmentation method that uses category-independent object proposals as its basic element and can extract multiple foreground objects in a video set. The use of object elements overcomes limitations of low-level feature representations in separating complex foregrounds and backgrounds. We formulate object-based co-segmentation as a co-selection graph in which regions with foreground-like characteristics are favored while also accounting for intra-video and inter-video foreground coherence. To handle multiple foreground objects, we expand the co-selection graph model into a proposed multi-state selection graph model (MSG) that optimizes the segmentations of different objects jointly. This extension into the MSG can be applied not only to our co-selection graph, but also can be used to turn any standard graph model into a multi-state selection solution that can be optimized directly by the existing energy minimization techniques. Our experiments show that our object-based multiple foreground video co-segmentation method (ObMiC) compares well to related techniques on both single and multiple foreground cases.

Social image‐sharing websites have attracted a large number of users. These systems allow users to associate geolocation information with their images, which is essential for many interesting applications. However, only a small fraction of social images have geolocation information. Thus, an automated tool for suggesting geolocation is essential to help users geotag their images. In this article, we use a large data set consisting of 221 million Flickr images uploaded by 2.2 million users. For the first time, we analyze user uploading patterns, user geotagging behaviors, and the relationship between the taken‐time gap and the geographical distance between two images from the same user. Based on the findings, we represent a user profile by historical tags for the user and build a multinomial model on the user profile for geotagging. We further propose a unified framework to suggest geolocations for images, which combines the information from both image tags and the user profile. Experimental results show that for images uploaded by users who have never done geotagging, our method outperforms the state‐of‐the‐art method by 10.6 to 34.2%, depending on the granularity of the prediction. For images from users who have done geotagging, a simple method is able to achieve very high accuracy.

In this paper, we study the heterogeneous domain adaptation (HDA) problem, in which the data from the source domain and the target domain are represented by heterogeneous features with different dimensions. By introducing two different projection matrices, we first transform the data from two domains into a common subspace such that the similarity between samples across different domains can be measured. We then propose a new feature mapping function for each domain, which augments the transformed samples with their original features and zeros. Existing supervised learning methods (e.g., SVM and SVR) can be readily employed by incorporating our newly proposed augmented feature representations for supervised HDA. As a showcase, we propose a novel method called Heterogeneous Feature Augmentation (HFA) based on SVM. We show that the proposed formulation can be equivalently derived as a standard Multiple Kernel Learning (MKL) problem, which is convex and thus the global solution can be guaranteed. To additionally utilize the unlabeled data in the target domain, we further propose the semi-supervised HFA (SHFA) which can simultaneously learn the target classifier as well as infer the labels of unlabeled target samples. Comprehensive experiments on three different applications clearly demonstrate that our SHFA and HFA outperform the existing HDA methods.

Hashing methods that aim to seek a compact binary code for each image are demonstrated to be efficient for scalable content-based image retrieval. In this paper, we propose a new hashing method called semisupervised kernel hyperplane learning (SKHL) for semantic image retrieval by modeling each hashing function as a nonlinear kernel hyperplane constructed from an unlabeled dataset. Moreover, a Fisher-like criterion is proposed to learn the optimal kernel hyperplanes and hashing functions, using only weakly labeled training samples with side information. To further integrate different types of features, we also incorporate multiple kernel learning (MKL) into the proposed SKHL (called SKHL-MKL), leading to better hashing functions. Comprehensive experiments on CIFAR-100 and NUS-WIDE datasets demonstrate the effectiveness of our SKHL and SKHL-MKL.

We propose a new graph based hashing method called spectral embedded hashing (SEH) for large-scale image retrieval. We first introduce a new regularizer into the objective function of the recent work spectral hashing to control the mismatch between the resultant hamming embedding and the low-dimensional data representation, which is obtained by using a linear regression function. This linear regression function can be employed to effectively handle the out-of-sample data, and the introduction of the new regularizer makes SEH better cope with the data sampled from a nonlinear manifold. Considering that SEH cannot efficiently cope with the high dimensional data, we further extend SEH to kernel SEH (KSEH) to improve the efficiency and effectiveness, in which a nonlinear regression function can also be employed to obtain the low dimensional data representation. We also develop a new method to efficiently solve the approximate solution for the eigenvalue decomposition problem in SEH and KSEH. Moreover, we show that some existing hashing methods are special cases of our KSEH. Our comprehensive experiments on CIFAR, Tiny-580K, NUS-WIDE, and Caltech-256 datasets clearly demonstrate the effectiveness of our methods.

Domain adaptation has shown promising results in computer vision applications. In this paper, we propose a new unsupervised domain adaptation method called domain adaptation by shifting covariance (DASC) for object recognition without requiring any labeled samples from the target domain. By characterizing samples from each domain as one covariance matrix, the source and target domain are represented into two distinct points residing on a Riemannian manifold. Along the geodesic constructed from the two points, we then interpolate some intermediate points (i.e., covariance matrices), which are used to bridge the two domains. By utilizing the principal components of each covariance matrix, samples from each domain are further projected into intermediate feature spaces, which finally leads to domain-invariant features after the concatenation of these features from intermediate points. In the multiple source domain adaptation task, we also need to effectively integrate different types of features between each pair of source and target domains. We additionally propose an SVM based method to simultaneously learn the optimal target classifier as well as the optimal weights for different source domains. Extensive experiments demonstrate the effectiveness of our method for both single source and multiple source domain adaptation tasks.

Influence maximization is a fundamental research problem in social networks. Viral marketing, one of its applications, is to get a small number of users to adopt a product, which subsequently triggers a large cascade of further adoptions by utilizing “Word-of-Mouth” effect in social networks. Time plays an important role in the influence spread from one user to another and the time needed for a user to influence another varies. In this paper, we propose the time constrained influence maximization problem. We show that the problem is NP-hard, and prove the monotonicity and submodularity of the time constrained influence spread function. Based on this, we develop a greedy algorithm. To improve the algorithm scalability, we propose the concept of Influence Spreading Path in social networks and develop a set of new algorithms for the time constrained influence maximization problem. We further parallelize the algorithms for achieving more time savings. Additionally, we generalize the proposed algorithms for the conventional influence maximization problem without time constraints. All of the algorithms are evaluated over four public available datasets. The experimental results demonstrate the efficiency and effectiveness of the algorithms for both conventional influence maximization problem and its time constrained version.

Co-clustering is a commonly used technique for tapping the rich meta-information of multimedia web documents, including category, annotation, and description, for associative discovery. However, most co-clustering methods proposed for heterogeneous data do not consider the representation problem of short and noisy text and their performance is limited by the empirical weighting of the multi-modal features. In this paper, we propose a generalized form of Heterogeneous Fusion Adaptive Resonance Theory, called GHF-ART, for co-clustering of large-scale web multimedia documents. By extending the two-channel Heterogeneous Fusion ART (HF-ART) to multiple channels, GHF-ART is designed to handle multimedia data with an arbitrarily rich level of meta-information. For handling short and noisy text, GHF-ART does not learn directly from the textual features. Instead, it identifies key tags by learning the probabilistic distribution of tag occurrences. More importantly, GHF-ART incorporates an adaptive method for effective fusion of multi-modal features, which weights the features of multiple data sources by incrementally measuring the importance of feature modalities through the intra-cluster scatters. Extensive experiments on two web image data sets and one text document set have shown that GHF-ART achieves significantly better clustering performance and is much faster than many existing state-of-the-art algorithms.

Most existing superpixel algorithms ignore the spatial structure and regularity properties, which result in undesirable sizes and location relationships for the subsequent processing. In this paper, we introduce a new method to generate the regularity preserved superpixels. Starting from the lattice seeds, our method relocates them to the pixel with locally maximal edge magnitudes and treats them as the superpixel junctions. Then, the shortest path algorithm is employed to find the local optimal boundary connecting each adjacent junction pair. Thanks to the local constraints, our method obtains homogeneous superpixels with adjacency in lowly textured and uniform regions and simultaneously preserves the boundary adherence in the high contrast contents. Our method preserves the regularity property without significantly sacrificing the segmentation accuracy. Moreover, we extend this regular constraint for generating the supervoxels. Our method obtains the regular supervoxels, which preserves the structural relation on both spatial and temporal spaces of the video. Quantitative and qualitative experimental results on benchmark datasets demonstrate that our simple but effective method outperforms the existing regular superpixel methods.

In this work, we propose to leverage a large number of loosely labeled web videos (e.g., from YouTube) and web images (e.g., from Google/Bing image search) for visual event recognition in consumer videos without requiring any labeled consumer videos. We formulate this task as a new multi-domain adaptation problem with heterogeneous sources, in which the samples from different source domains can be represented by different types of features with different dimensions (e.g., the SIFT features from web images and space-time (ST) features from web videos) while the target domain samples have all types of features. To effectively cope with the heterogeneous sources where some source domains are more relevant to the target domain, we propose a new method called Multi-domain Adaptation with Heterogeneous Sources (MDA-HS) to learn an optimal target classifier, in which we simultaneously seek the optimal weights for different source domains with different types of features as well as infer the labels of unlabeled target domain data based on multiple types of features. We solve our optimization problem by using the cutting-plane algorithm based on group based multiple kernel learning. Comprehensive experiments on two datasets demonstrate the effectiveness of MDA-HS for event recognition in consumer videos.

In many real-world face recognition scenarios, face images can hardly be aligned accurately due to complex appearance variations or low-quality images. To address this issue, we propose a new approach to extract robust face region descriptors. Specifically, we divide each image (resp. video) into several spatial blocks (resp. spatial-temporal volumes) and then represent each block (resp. volume) by sum-pooling the nonnegative sparse codes of position-free patches sampled within the block (resp. volume). Whitened Principal Component Analysis (WPCA) is further utilized to reduce the feature dimension, which leads to our Spatial Face Region Descriptor (SFRD) (resp. Spatial-Temporal Face Region Descriptor, STFRD) for images (resp. videos). Moreover, we develop a new distance metric learning method for face verification called Pairwise-constrained Multiple Metric Learning (PMML) to effectively integrate the face region descriptors of all blocks (resp. volumes) from an image (resp. a video). Our work achieves the state-of-the-art performances on two real-world datasets LFW and YouTube Faces (YTF) according to the restricted protocol.

We study in this paper the problem of learning classifiers from ambiguously labeled images. For instance, in the collection of new images, each image contains some samples of interest (\emph{e.g.,} human faces), and its associated caption has labels with the true ones included, while the sample-label association is unknown. The task is to learn classifiers from these ambiguously labeled images and generalize to new images. An essential consideration here is how to make use of the information embedded in the relations between samples and labels, both within each image and across the image set. To this end, we propose a novel framework to address this problem. Our framework is motivated by the observation that samples from the same class repetitively appear in the collection of ambiguously labeled training images, while they are just ambiguously labeled in each image. If we can identify samples of the same class from each image and associate them across the image set, the matrix formed by the samples from the same class would be ideally low-rank. By leveraging such a low-rank assumption, we can simultaneously optimize a partial permutation matrix (PPM) for each image, which is formulated in order to exploit all information between samples and labels in a principled way. The obtained PPMs can be readily used to assign labels to samples in training images, and then a standard SVM classifier can be trained and used for unseen data. Experiments on benchmark datasets show the effectiveness of our proposed method.

We present a method for estimating human scan paths, which are sequences of gaze shifts that follow visual attention over an image. In this work, scan paths are modeled based on three principal factors that influence human attention, namely low-level feature saliency, spatial position, and semantic content. Low-level feature saliency is formulated as transition probabilities between different image regions based on feature differences. The effect of spatial position on gaze shifts is modeled as a Levy flight with the shifts following a 2D Cauchy distribution. To account for semantic content, we propose to use a Hidden Markov Model (HMM) with a Bag-of-Visual-Words descriptor of image regions. An HMM is well-suited for this purpose in that 1) the hidden states, obtained by unsupervised learning, can represent latent semantic concepts, 2) the prior distribution of the hidden states describes visual attraction to the semantic concepts, and 3) the transition probabilities represent human gaze shift patterns. The proposed method is applied to task-driven viewing processes. Experiments and analysis performed on human eye gaze data verify the effectiveness of this method.

BACKGROUND: Computer-aided diagnosis for screening utilizes computer-based analytical methodologies to process patient information. Glaucoma is the leading irreversible cause of blindness. Due to the lack of an effective and standard screening practice, more than 50% of the cases are undiagnosed, which prevents the early treatment of the disease. OBJECTIVE: To design an automatic glaucoma diagnosis architecture automatic glaucoma diagnosis through medical imaging informatics (AGLAIA-MII) that combines patient personal data, medical retinal fundus image, and patient's genome information for screening. MATERIALS AND METHODS: 2258 cases from a population study were used to evaluate the screening software. These cases were attributed with patient personal data, retinal images and quality controlled genome data. Utilizing the multiple kernel learning-based classifier, AGLAIA-MII, combined patient personal data, major image features, and important genome single nucleotide polymorphism (SNP) features. RESULTS AND DISCUSSION: Receiver operating characteristic curves were plotted to compare AGLAIA-MII's performance with classifiers using patient personal data, images, and genome SNP separately. AGLAIA-MII was able to achieve an area under curve value of 0.866, better than 0.551, 0.722 and 0.810 by the individual personal data, image and genome information components, respectively. AGLAIA-MII also demonstrated a substantial improvement over the current glaucoma screening approach based on intraocular pressure. CONCLUSIONS: AGLAIA-MII demonstrates for the first time the capability of integrating patients' personal data, medical retinal image and genome information for automatic glaucoma diagnosis and screening in a large dataset from a population study. It paves the way for a holistic approach for automatic objective glaucoma diagnosis and screening.

We first propose a new low-level visual feature, called spatio-temporal context distribution feature of interest points, to describe human actions. Each action video is expressed as a set of relative XYT coordinates between pairwise interest points in a local region. We learn a global Gaussian mixture model (GMM) (referred to as a universal background model) using the relative coordinate features from all the training videos, and then we represent each video as the normalized parameters of a video-specific GMM adapted from the global GMM. In order to capture the spatio-temporal relationships at different levels, multiple GMMs are utilized to describe the context distributions of interest points over multiscale local regions. Motivated by the observation that some actions share similar motion patterns, we additionally propose a novel mid-level class correlation feature to capture the semantic correlations between different action classes. Each input action video is represented by a set of decision values obtained from the pre-learned classifiers of all the action classes, with each decision value measuring the likelihood that the input video belongs to the corresponding action class. Moreover, human actions are often associated with some specific natural environments and also exhibit high correlation with particular scene classes. It is therefore beneficial to utilize the contextual scene information for action recognition. In this paper, we build the high-level co-occurrence relationship between action classes and scene classes to discover the mutual contextual constraints between action and scene. By treating the scene class label as a latent variable, we propose to use the latent structural SVM (LSSVM) model to jointly capture the compatibility between multilevel action features (e.g., low-level visual context distribution feature and the corresponding mid-level class correlation feature) and action classes, the compatibility between multilevel scene features (i.e., SIFT feature and the corresponding class correlation feature) and scene classes, and the contextual relationship between action classes and scene classes. Extensive experiments on UCF Sports, YouTube and UCF50 datasets demonstrate the effectiveness of the proposed multilevel features and action-scene interaction based LSSVM model for human action recognition. Moreover, our method generally achieves higher recognition accuracy than other state-of-the-art methods on these datasets.

With the invention of the low-cost Microsoft Kinect sensor, high-resolution depth and visual (RGB) sensing has become available for widespread use. The complementary nature of the depth and visual information provided by the Kinect sensor opens up new opportunities to solve fundamental problems in computer vision. This paper presents a comprehensive review of recent Kinect-based computer vision algorithms and applications. The reviewed approaches are classified according to the type of vision problems that can be addressed or enhanced by means of the Kinect sensor. The covered topics include preprocessing, object tracking and recognition, human activity analysis, hand gesture analysis, and indoor 3-D mapping. For each category of methods, we outline their main algorithmic contributions and summarize their advantages/differences compared to their RGB counterparts. Finally, we give an overview of the challenges in this field and future research trends. This paper is expected to serve as a tutorial and source of references for Kinect-based computer vision researchers.

Kinect sensor, high-resolution depth and visual (RGB) sensing has become available for widespread use as an off-the-shelf technology. This special issue is specifically dedicated to new algorithms and/or new applications based on the Kinect (or similar RGB-D) sensors. In total, we received over ninety submissions from more than twenty countries all around the world. The submissions cover a wide range of areas including object and scene classification, 3-D pose estimation, visual tracking, data fusion, human action/activity recognition, 3-D reconstruction, mobile robotics, and so on. After two rounds of review by at least two (mostly three) expert reviewers for each paper, the Guest Editors have selected twelve high-quality papers to be included in this highly popular special issue. The papers that comprise this issue are briefly summarized.

The bag-of-words (BoW) model has been known as an effective method for large-scale image search and indexing. Recent work shows that the performance of the model can be further improved by using the embedding method. While different variants of the BoW model and embedding method have been developed, less effort has been made to discover their underlying working mechanism. In this paper, we systematically investigate the image search performance variation with respect to a few factors of the BoW model, and study how to employ the embedding method to further improve the image search performance. Subsequently, we summarize several observations based on the experiments on descriptor matching. To validate these observations in a real image search, we propose an effective and efficient image search scheme, in which the BoW model and embedding method are jointly optimized in terms of effectiveness and efficiency by following these observations. Our comprehensive experiments demonstrate that it is beneficial to employ these observations to develop an image search algorithm, and the proposed image search scheme outperforms state-of-the art methods in both effectiveness and efficiency.

In this paper, we propose a new scheme called Prototype Hyperplane Learning (PHL) for face verification in the wild using only weakly labeled training samples (i.e., we only know whether each pair of samples are from the same class or different classes without knowing the class label of each sample) by leveraging a large number of unlabeled samples in a generic data set. Our scheme represents each sample in the weakly labeled data set as a mid-level feature with each entry as the corresponding decision value from the classification hyperplane (referred to as the prototype hyperplane) of one Support Vector Machine (SVM) model, in which a sparse set of support vectors is selected from the unlabeled generic data set based on the learnt combination coefficients. To learn the optimal prototype hyperplanes for the extraction of mid-level features, we propose a Fisher’s Linear Discriminant-like (FLD-like) objective function by maximizing the discriminability on the weakly labeled data set with a constraint enforcing sparsity on the combination coefficients of each SVM model, which is solved by using an alternating optimization method. Then, we use the recent work called Side-Information based Linear Discriminant (SILD) analysis for dimensionality reduction and a cosine similarity measure for final face verification. Comprehensive experiments on two data sets, Labeled Faces in the Wild (LFW) and YouTube Faces, demonstrate the effectiveness of our scheme.

The one-to-one correspondence between co-occurrence image patches of two different resolutions is extensively used in example-based super-resolution (SR). Due to the dimensionality gap between low resolution (LR) and high resolution (HR) spaces, however, an LR patch may correspond to a number of HR patches in practice. This ambiguity is difficult to be overcome with examples representing a deterministic mapping. In this paper, we propose a statistical method for exploiting the one-to-many correspondence between LR and HR patches, which we call soft information and decision. Soft information means an LR patch is mapped to a pixel-wise distribution of all its possible HR counterparts, rather than a single or a limited set of HR candidates. Relying on the soft information, example-based SR is then regarded as an optimization problem to best preserve the local consistency in the recovered HR image. This problem is solved with an efficient message passing algorithm with a factor graph model. The final decision on the HR pixel value is made upon the maximum a posteriori estimation and is called a soft decision. Experimental results demonstrate the superiority of the proposed method compared with the state-of-the-art methods, in terms of both the subjective and objective quality of synthesized HR images.

This issue marks the first anniversary issue of IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS after it changed its name from IEEE TRANSACTIONS ON NEURAL NETWORKS. I am happy to report that we had a great year! The number of new submissions in a year exceeded 1,000 for the first time in the history of TNN/TNNLS. IEEE TNN had a very successful development for 22 years from 1990 to 2011, and we have good reasons to believe that IEEE TNNLS will have many more years of successful growth.

Multiple kernel learning (MKL) has been proposed for kernel methods by learning the optimal kernel from a set of predefined base kernels. However, the traditional L1MKL method often achieves worse results than the simplest method using the average of base kernels (i.e., average kernel) in some practical applications. In order to improve the effectiveness of MKL, this paper presents a novel soft margin perspective for MKL. Specifically, we introduce an additional slack variable called kernel slack variable to each quadratic constraint of MKL, which corresponds to one support vector machine model using a single base kernel. We first show that L1MKL can be deemed as hard margin MKL, and then we propose a novel soft margin framework for MKL. Three commonly used loss functions, including the hinge loss, the square hinge loss, and the square loss, can be readily incorporated into this framework, leading to the new soft margin MKL objective functions. Many existing MKL methods can be shown as special cases under our soft margin framework. For example, the hinge loss soft margin MKL leads to a new box constraint for kernel combination coefficients. Using different hyper-parameter values for this formulation, we can inherently bridge the method using average kernel, L1MKL, and the hinge loss soft margin MKL. The square hinge loss soft margin MKL unifies the family of elastic net constraint/regularizer based approaches; and the square loss soft margin MKL incorporates L2MKL naturally. Moreover, we also develop efficient algorithms for solving both the hinge loss and square hinge loss soft margin MKL. Comprehensive experimental studies for various MKL algorithms on several benchmark data sets and two real world applications, including video action recognition and event recognition demonstrate that our proposed algorithms can efficiently achieve an effective yet sparse solution for MKL.

Conflict resolution is one major challenge in real-time distributed collaborative 3D design systems, which allow concurrent collaborative work on shared 3D documents. Operational Transformation (OT) is a core conflict resolution technique in a range of real-world collaborative systems. No existing OT technique is, however, capable of resolving conflicts among objects with dependency relations, i.e. an update of one object may propagate to other connected/dependent objects, which is commonly used in 3D or complex graphic design systems. This paper contributes a novel OT solution with such a capability for collaborative 3D design systems. This work is the first to extend OT capability to dependency conflict resolution, and OT application scope from 1D/2D to 3D applications, thus advancing the state-of-the-art of OT in both theory and practical application. The proposed solution has been theoretically verified for its correctness in detecting dependency conflicts and achieving consistent results, and implemented in the CoMaya collaborative design system.

Recent work has demonstrated the effectiveness of domain adaptation methods for computer vision applications. In this work, we propose a new multiple source domain adaptation method called Domain Selection Machine (DSM) for event recognition in consumer videos by leveraging a large number of loosely labeled web images from different sources (e.g., Flickr.com and Photosig.com), in which there are no labeled consumer videos. Specifically, we first train a set of SVM classifiers (referred to as source classifiers) by using the SIFT features of web images from different source domains. We propose a new parametric target decision function to effectively integrate the static SIFT features from web images/video keyframes and the spacetime (ST) features from consumer videos. In order to select the most relevant source domains, we further introduce a new data-dependent regularizer into the objective of Support Vector Regression (SVR) using the ∊-insensitive loss, which enforces the target classifier shares similar decision values on the unlabeled consumer videos with the selected source classifiers. Moreover, we develop an alternating optimization algorithm to iteratively solve the target decision function and a domain selection vector which indicates the most relevant source domains. Extensive experiments on three real-world datasets demonstrate the effectiveness of our proposed method DSM over the state-of-the-art by a performance gain up to 46.41%.

Multiple Instance Learning (MIL) has been widely exploited in many computer vision tasks, such as image retrieval, object tracking and so on. To handle ambiguity of instance labels in positive bags, the training process of traditional MIL methods is usually computationally expensive, which limits the applications of MIL in more computer vision tasks. In this paper, we propose a novel batch mode framework, namely Batch mode Adaptive Multiple Instance Learning (BAMIL), to accelerate the instance-level MIL methods. Specifically, instead of using all training bags at once, we divide the training bags into several sets of bags (i.e., batches). At each time, we use one batch of training bags to train a new classifier which is adapted from the latest pre-learned classifier. Such batch mode framework significantly accelerates the traditional MIL methods for large scale applications and can be also used in dynamic environments such as object tracking. The experimental results show that our BAMIL is much faster than the recently developed MIL with constrained positive bags while achieves comparable performance for text-based web image retrieval. In dynamic settings, BAMIL also achieves the better overall performance for object tracking when compared with other online MIL methods.

We propose a multi-view learning approach called co-labeling which is applicable for several machine learning problems where the labels of training samples are uncertain, including semi-supervised learning (SSL), multi-instance learning (MIL) and max-margin clustering (MMC). Particularly, we first unify those problems into a general ambiguous problem in which we simultaneously learn a robust classifier as well as find the optimal training labels from a finite label candidate set. To effectively utilize multiple views of data, we then develop our co-labeling approach for the general multi-view ambiguous problem. In our work, classifiers trained on different views can teach each other by iteratively passing the predictions of training samples from one classifier to the others. The predictions from one classifier are considered as label candidates for the other classifiers. To train a classifier with a label candidate set for each view, we adopt the Multiple Kernel Learning (MKL) technique by constructing the base kernel through associating the input kernel calculated from input features with one label candidate. Compared with the traditional co-training method which was specifically designed for SSL, the advantages of our co-labeling are two-fold: 1) it can be applied to other ambiguous problems such as MIL and MMC, 2) it is more robust by using the MKL method to integrate multiple labeling candidates obtained from different iterations and biases. Promising results on several real-world multi-view data sets clearly demonstrate the effectiveness of our proposed co-labeling for both MIL and SSL.

Influence maximization is a fundamental research problem in social networks. Viral marketing, one of its applications, is to get a small number of users to adopt a product, which subsequently triggers a large cascade of further adoptions by utilizing "Word-of-Mouth" effect in social networks. Influence maximization problem has been extensively studied recently. However, none of the previous work considers the time constraint in the influence maximization problem. In this paper, we propose the time constrained influence maximization problem. We show that the problem is NP-hard, and prove the monotonicity and submodularity of the time constrained influence spread function. Based on this, we develop a greedy algorithm with performance guarantees. To improve the algorithm scalability, we propose two Influence Spreading Path based methods. Extensive experiments conducted over four public available datasets demonstrate the efficiency and effectiveness of the Influence Spreading Path based methods.

Data ambiguities exist in many data mining and machine learning applications such as text categorization and image retrieval. For instance, it is generally beneficial to utilize the ambiguous unlabeled documents to learn a more robust classifier for text categorization under the semi-supervised learning setting. To handle general data ambiguities, we present a unified kernel learning framework named Input-Output Kernel Learning (IOKL). Based on our framework, we further propose a novel soft margin group sparse Multiple Kernel Learning (MKL) formulation by introducing a group kernel slack variable to each group of base input-output kernels. Moreover, an efficient block-wise coordinate descent algorithm with an analytical solution for the kernel combination coefficients is developed to solve the proposed formulation. We conduct comprehensive experiments on benchmark datasets for both semi-supervised learning and multiple instance learning tasks, and also apply our IOKL framework to a computer vision application called text-based image retrieval on the NUS-WIDE dataset. Promising results demonstrate the effectiveness of our proposed IOKL framework.

Cross-domain learning methods have shown promising results by leveraging labeled patterns from the auxiliary domain to learn a robust classifier for the target domain which has only a limited number of labeled samples. To cope with the considerable change between feature distributions of different domains, we propose a new cross-domain kernel learning framework into which many existing kernel methods can be readily incorporated. Our framework, referred to as Domain Transfer Multiple Kernel Learning (DTMKL), simultaneously learns a kernel function and a robust classifier by minimizing both the structural risk functional and the distribution mismatch between the labeled and unlabeled samples from the auxiliary and target domains. Under the DTMKL framework, we also propose two novel methods by using SVM and prelearned classifiers, respectively. Comprehensive experiments on three domain adaptation data sets (i.e., TRECVID, 20 Newsgroups, and email spam data sets) demonstrate that DTMKL-based methods outperform existing cross-domain learning and multiple kernel learning methods.

We propose a visual event recognition framework for consumer videos by leveraging a large amount of loosely labeled web videos (e.g., from YouTube). Observing that consumer videos generally contain large intraclass variations within the same type of events, we first propose a new method, called Aligned Space-Time Pyramid Matching (ASTPM), to measure the distance between any two video clips. Second, we propose a new transfer learning method, referred to as Adaptive Multiple Kernel Learning (A-MKL), in order to 1) fuse the information from multiple pyramid levels and features (i.e., space-time features and static SIFT features) and 2) cope with the considerable variation in feature distributions between videos from two domains (i.e., web video domain and consumer video domain). For each pyramid level and each type of local features, we first train a set of SVM classifiers based on the combined training set from two domains by using multiple base kernels from different kernel types and parameters, which are then fused with equal weights to obtain a prelearned average classifier. In A-MKL, for each event class we learn an adapted target classifier based on multiple base kernels and the prelearned average classifiers from this event class or all the event classes by minimizing both the structural risk functional and the mismatch between data distributions of two domains. Extensive experiments demonstrate the effectiveness of our proposed framework that requires only a small number of labeled consumer videos by leveraging web data. We also conduct an in-depth investigation on various aspects of the proposed method A-MKL, such as the analysis on the combination coefficients on the prelearned classifiers, the convergence of the learning algorithm, and the performance variation by using different proportions of labeled consumer videos. Moreover, we show that A-MKL using the prelearned classifiers from all the event classes leads to better performance when compared with A-MK- using the prelearned classifiers only from each individual event class.

We present a new framework for multimedia content analysis and retrieval which consists of two independent algorithms. First, we propose a new semi-supervised algorithm called ranking with Local Regression and Global Alignment (LRGA) to learn a robust Laplacian matrix for data ranking. In LRGA, for each data point, a local linear regression model is used to predict the ranking scores of its neighboring points. A unified objective function is then proposed to globally align the local models from all the data points so that an optimal ranking score can be assigned to each data point. Second, we propose a semi-supervised long-term Relevance Feedback (RF) algorithm to refine the multimedia data representation. The proposed long-term RF algorithm utilizes both the multimedia data distribution in multimedia feature space and the history RF information provided by users. A trace ratio optimization problem is then formulated and solved by an efficient algorithm. The algorithms have been applied to several content-based multimedia retrieval applications, including cross-media retrieval, image retrieval, and 3D motion/pose data retrieval. Comprehensive experiments on four data sets have demonstrated its advantages in precision, robustness, scalability, and computational efficiency.

We propose a new semisupervised learning algorithm, referred to as patch distribution compatible semisupervised dimension reduction, for face and human gait recognition. Each image (a face image or an average human silhouette image) is first represented as a set of local patch features and it is further characterized as the corresponding patch distribution feature, which can be expressed as an image-specific Gaussian mixture model (GMM) adapted from the universal background model. Assuming that the individual components of the image-specific GMMs from all the training images reside on a submanifold, we assign a component-level prediction label matrix to each individual GMM component and introduce a new regularizer based on a set of local submanifold smoothness assumptions in our objective function. We also constrain each component-level prediction label matrix to be consistent with the image-level prediction label matrix , as well as enforce to be close to the given labels for the labeled samples. We further use a linear regression function to provide embeddings for the training data and the unseen test data. Inspired by the recent work flexible manifold embedding, we additionally integrate the regression residue in our objective function to measure the mismatch between and , such that our method can better cope with the data sampled from a nonlinear manifold. Finally, the optimal solutions of the component-level prediction label matrix , the image-level prediction label matrix , the projection matrix , and the bias term b can be simultaneously obtained. Comprehensive experiments on three benchmark face databases CMU PIE, FERET, and AR as well as the USF HumanID gait database clearly demonstrate the effectiveness of our algorithm over other state-of-the-art semisupervised dimension reduction methods.

Histograms have been widely used for feature representation in image and video content analysis. However, due to the orderless nature of the summarization process, histograms generally lack spatial information. This may degrade their discrimination capability in visual classification tasks. Although there have been several research attempts to encode spatial context into histograms, how to extend the encodings to higher order spatial context is still an open problem. In this paper,we propose a general histogram contextualization method to encode efficiently higher order spatial context. The method is based on the cooccurrence of local visual homogeneity patterns and hence is able to generate more discriminative histogram representations while remaining compact and robust. Moreover, we also investigate how to extend the histogram contextualization to multiple modalities of context. It is shown that the proposed method can be naturally extended to combine both temporal and spatial context and facilitate video content analysis. In addition, a method to combine cross-feature context with spatial context via the technique of random forest is also introduced in this paper. Comprehensive experiments on face image classification and human activity recognition tasks demonstrate the superiority of the proposed histogram contextualization method compared with the existing encoding methods.

Visual saliency is a useful clue to depict visually important image/video contents in many multimedia applications. In visual saliency estimation, a feasible solution is to learn a "feature-saliency" mapping model from the user data obtained by manually labeling activities or eye-tracking devices. However, label ambiguities may also arise due to the inaccurate and inadequate user data. To process the noisy training data, we propose a multi-instance learning to rank approach for visual saliency estimation. In our approach, the correlations between various image patches are incorporated into an ordinal regression framework. By iteratively refining a ranking model and relabeling the image patches with respect to their mutual correlations, the label ambiguities can be effectively removed from the training data. Consequently, visual saliency can be effectively estimated by the ranking model, which can pop out real targets and suppress real distractors. Extensive experiments on two public image data sets show that our approach outperforms 11 state-of-the-art methods remarkably in visual saliency estimation.

In this paper, we propose a new patch distribution feature (PDF) (i.e., referred to as Gabor-PDF) for human gait recognition. We represent each gait energy image (GEI) as a set of local augmented Gabor features, which concatenate the Gabor features extracted from different scales and different orientations together with the X-Y coordinates. We learn a global Gaussian mixture model (GMM) (i.e., referred to as the universal background model) with the local augmented Gabor features from all the gallery GEIs; then, each gallery or probe GEI is further expressed as the normalized parameters of an image-specific GMM adapted from the global GMM. Observing that one video is naturally represented as a group of GEIs, we also propose a new classification method called locality-constrained group sparse representation (LGSR) to classify each probe video by minimizing the weighted l(1, 2) mixed-norm-regularized reconstruction error with respect to the gallery videos. In contrast to the standard group sparse representation method that is a special case of LGSR, the group sparsity and local smooth sparsity constraints are both enforced in LGSR. Our comprehensive experiments on the benchmark USF HumanID database demonstrate the effectiveness of the newly proposed feature Gabor-PDF and the new classification method LGSR for human gait recognition. Moreover, LGSR using the new feature Gabor-PDF achieves the best average Rank-1 and Rank-5 recognition rates on this database among all gait recognition algorithms proposed to date.

In this paper, we propose a new tag-based image retrieval framework to improve the retrieval performance of a group of related personal images captured by the same user within a short period of an event by leveraging millions of training web images and their associated rich textual descriptions. For any given query tag (e.g., “car”), the inverted file method is employed to automatically determine the relevant training web images that are associated with the query tag and the irrelevant training web images that are not associated with the query tag. Using these relevant and irrelevant web images as positive and negative training data respectively, we propose a new classification method called support vector machine (SVM) with augmented features (AFSVM) to learn an adapted classifier by leveraging the prelearned SVM classifiers of popular tags that are associated with a large number of relevant training web images. Treating the decision values of one group of test photos from AFSVM classifiers as the initial relevance scores, in the subsequent group-based refinement process, we propose to use the Laplacian regularized least squares method to further refine the relevance scores of test photos by utilizing the visual similarity of the images within the group. Based on the refined relevance scores, our proposed framework can be readily applied to tag-based image retrieval for a group of raw consumer photos without any textual descriptions or a group of Flickr photos with noisy tags. Moreover, we propose a new method to better calculate the relevance scores for Flickr photos. Extensive experiments on two datasets demonstrate the effectiveness of our framework.

Semi-supervised learning (SSL), as a powerful tool to learn from a limited number of labeled data and a large number of unlabeled data, has been attracting increasing attention in the machine learning community. In particular, the manifold regularization framework has laid solid theoretical foundations for a large family of SSL algorithms, such as Laplacian support vector machine (LapSVM) and Laplacian regularized least squares (LapRLS). However, most of these algorithms are limited to small scale problems due to the high computational cost of the matrix inversion operation involved in the optimization problem. In this paper, we propose a novel framework called Laplacian embedded regression by introducing an intermediate decision variable into the manifold regularization framework. By using ∈-insensitive loss, we obtain the Laplacian embedded support vector regression (LapESVR) algorithm, which inherits the sparse solution from SVR. Also, we derive Laplacian embedded RLS (LapERLS) corresponding to RLS under the proposed framework. Both LapESVR and LapERLS possess a simpler form of a transformed kernel, which is the summation of the original kernel and a graph kernel that captures the manifold structure. The benefits of the transformed kernel are two-fold: (1) we can deal with the original kernel matrix and the graph Laplacian matrix in the graph kernel separately and (2) if the graph Laplacian matrix is sparse, we only need to perform the inverse operation for a sparse matrix, which is much more efficient when compared with that for a dense one. Inspired by kernel principal component analysis, we further propose to project the introduced decision variable into a subspace spanned by a few eigenvectors of the graph Laplacian matrix in order to better reflect the data manifold, as well as accelerate the calculation of the graph kernel, allowing our methods to efficiently and effectively cope with large scale SSL problems. Extensive experiments on both toy and real world data sets show the effectiveness and scalability of the proposed framework.

In this paper, we propose a new framework called domain adaptation machine (DAM) for the multiple source domain adaption problem. Under this framework, we learn a robust decision function (referred to as target classifier) for label prediction of instances from the target domain by leveraging a set of base classifiers which are prelearned by using labeled instances either from the source domains or from the source domains and the target domain. With the base classifiers, we propose a new domain-dependent regularizer based on smoothness assumption, which enforces that the target classifier shares similar decision values with the relevant base classifiers on the unlabeled instances from the target domain. This newly proposed regularizer can be readily incorporated into many kernel methods (e.g., support vector machines (SVM), support vector regression, and least-squares SVM (LS-SVM)). For domain adaptation, we also develop two new domain adaptation methods referred to as FastDAM and UniverDAM. In FastDAM, we introduce our proposed domain-dependent regularizer into LS-SVM as well as employ a sparsity regularizer to learn a sparse target classifier with the support vectors only from the target domain, which thus makes the label prediction on any test instance very fast. In UniverDAM, we additionally make use of the instances from the source domains as Universum to further enhance the generalization ability of the target classifier. We evaluate our two methods on the challenging TRECIVD 2005 dataset for the large-scale video concept detection task as well as on the 20 newsgroups and email spam datasets for document retrieval. Comprehensive experiments demonstrate that FastDAM and UniverDAM outperform the existing multiple source domain adaptation methods for the two applications.

In this brief, we address the trace ratio (TR) problem for semi-supervised dimension reduction. We first reformulate the objective function of the recent work semi-supervised discriminant analysis (SDA) in a TR form. We also observe that in SDA the low-dimensional data representation F is constrained to be in the linear subspace spanned by the training data matrix X (i.e., F = X(T) W). In order to relax this hard constraint, we introduce a flexible regularizer ||F - X(T) W||(2) which models the regression residual into the reformulated objective function. With such relaxation, our method referred to as TR based flexible SDA (TR-FSDA) can better cope with data sampled from a certain type of nonlinear manifold that is somewhat close to a linear subspace. In order to address the non-trivial optimization problem in TR-FSDA, we further develop an iterative algorithm to simultaneously solve for the low-dimensional data representation F and the projection matrix W. Moreover, we theoretically prove that our iterative algorithm converges to the optimum based on the Newton-Raphson method. The experiments on two face databases, one shape image database and one webpage database demonstrate that TR-FSDA outperforms the existing semi-supervised dimension reduction methods.

The results of traditional clustering methods are usually unreliable as there is not any guidance from the data labels, while the class labels can be predicted more reliable by the semisupervised learning if the labels of partial data are given. In this paper, we propose an actively self-training clustering method, in which the samples are actively selected as training set to minimize an estimated Bayes error, and then explore semisupervised learning to perform clustering. Traditional graph-based semisupervised learning methods are not convenient to estimate the Bayes error; we develop a specific regularization framework on graph to perform semisupervised learning, in which the Bayes error can be effectively estimated. In addition, the proposed clustering algorithm can be readily applied in a semisupervised setting with partial class labels. Experimental results on toy data and real-world data sets demonstrate the effectiveness of the proposed clustering method on the unsupervised and the semisupervised setting. It is worthy noting that the proposed clustering method is free of initialization, while traditional clustering methods are usually dependent on initialization.

In this paper, we study the problem of detecting sudden pedestrian crossings to assist drivers in avoiding accidents. This application has two major requirements: to detect crossing pedestrians as early as possible just as they enter the view of the car-mounted camera and to maintain a false alarm rate as low as possible for practical purposes. Although many current sliding-window-based approaches using various features and classification algorithms have been proposed for image-/video-based pedestrian detection, their performance in terms of accuracy and processing speed falls far short of practical application requirements. To address this problem, we propose a three-level coarse-to-fine video-based framework that detects partially visible pedestrians just as they enter the camera view, with low false alarm rate and high speed. The framework is tested on a new collection of high-resolution videos captured from a moving vehicle and yields a performance better than that of state-of-the-art pedestrian detection while running at a frame rate of 55 fps.

In recent years, face recognition in the unconstrained environment has attracted increasing attentions, and a few methods have been evaluated on the Labeled Faces in the Wild (LFW) database. In the unconstrained conditions, sometimes we cannot obtain the full class label information of all the subjects. Instead we can only get the weak label information, such as the side-information, i.e., the image pairs from the same or different subjects. In this scenario, many multi-class methods (e.g., the well-known Fisher Linear Discriminant Analysis (FLDA)), fail to work due to the lack of full class label information. To effectively utilize the side-information in such case, we propose Side-Information based Linear Discriminant Analysis (SILD), in which the within-class and between-class scatter matrices are directly calculated by using the side-information. Moreover, we theoretically prove that our SILD method is equivalent to FLDA when the full class label information is available. Experiments on LFW and FRGC databases support our theoretical analysis, and SILD using multiple features also achieve promising performance when compared with the state-of-the-art methods.

We first propose a new spatio-temporal context distribution feature of interest points for human action recognition. Each action video is expressed as a set of relative XYT coordinates between pairwise interest points in a local region. We learn a global GMM (referred to as Universal Background Model, UBM) using the relative coordinate features from all the training videos, and then represent each video as the normalized parameters of a video-specific GMM adapted from the global GMM. In order to capture the spatio-temporal relationships at different levels, multiple GMMs are utilized to describe the context distributions of interest points over multi-scale local regions. To describe the appearance information of an action video, we also propose to use GMM to characterize the distribution of local appearance features from the cuboids centered around the interest points. Accordingly, an action video can be represented by two types of distribution features: 1) multiple GMM distributions of spatio-temporal context; 2) GMM distribution of local video appearance. To effectively fuse these two types of heterogeneous and complementary distribution features, we additionally propose a new learning algorithm, called Multiple Kernel Learning with Augmented Features (AFMKL), to learn an adapted classifier based on multiple kernels and the pre-learned classifiers of other action classes. Extensive experiments on KTH, multi-view IXMAS and complex UCF sports datasets demonstrate that our method generally achieves higher recognition accuracy than other state-of-the-art methods.

Relevant and irrelevant images collected from the Web (e.g., Flickr.com) have been employed as loosely labeled training data for image categorization and retrieval. In this work, we propose a new approach to learn a robust classifier for text-based image retrieval (TBIR) using relevant and irrelevant training web images, in which we explicitly handle noise in the loose labels of training images. Specifically, we first partition the relevant and irrelevant training web images into clusters. By treating each cluster as a "bag" and the images in each bag as "instances", we formulate this task as a multi-instance learning problem with constrained positive bags, in which each positive bag contains at least a portion of positive instances. We present a new algorithm called MIL-CPB to effectively exploit such constraints on positive bags and predict the labels of test instances (images). Observing that the constraints on positive bags may not always be satisfied in our application, we additionally propose a progressive scheme (referred to as Progressive MIL-CPB, or PMIL-CPB) to further improve the retrieval performance, in which we iteratively partition the top-ranked training web images from the current MIL-CPB classifier to construct more confident positive "bags "and then add these new "bags" as training data to learn the subsequent MIL-CPB classifiers. Comprehensive experiments on two challenging real-world web image data sets demonstrate the effectiveness of our approach. © 2011 IEEE.

In large-scale video database, lots of different videos frequently share the similar content copied from the same source. Generally, those videos have certain semantic correlations, such as being of similar events and sharing the same topic. Mining these semantic correlations can greatly facilitate video search. However, as a preprocessing step, detecting and localizing the copy pair among videos, i.e. copy detection problem, plays a key role for precise semantic mining. To meet the requirements in semantic mining scenario, we propose a frame fusion based copy detection scheme. In this scheme, the copy detection problem is converted to HMM decoding problem with three relaxed constraints, where Viterbi algorithm is employed to automatically detect the copy pair. The experimental results show that the proposed approach achieves high localization accuracy even when the copied clip undergoes some complex transformations, while achieving comparable performance compared with state-of-the-art copy detection methods.

The rapid popularization of digital cameras and mobile phone cameras has led to an explosive growth of personal photo collections by consumers. In this paper, we present a real-time textual query-based personal photo retrieval system by leveraging millions of Web images and their associated rich textual descriptions (captions, categories, etc.). After a user provides a textual query (e.g., “water”), our system exploits the inverted file to automatically find the positive Web images that are related to the textual query “water” as well as the negative Web images that are irrelevant to the textual query. Based on these automatically retrieved relevant and irrelevant Web images, we employ three simple but effective classification methods, k-Nearest Neighbor (kNN), decision stumps, and linear SVM, to rank personal photos. To further improve the photo retrieval performance, we propose two relevance feedback methods via cross-domain learning, which effectively utilize both the Web images and personal images. In particular, our proposed crossdomain learning methods can learn robust classifiers with only a very limited amount of labeled personal photos from the user by leveraging the prelearned linear SVM classifiers in real time. We further propose an incremental cross-domain learning method in order to significantly accelerate the relevance feedback process on large consumer photo databases. Extensive experiments on two consumer photo data sets demonstrate the effectiveness and efficiency of our system, which is also inherently not limited by any predefined lexicon.

Given a textual query in traditional text-based image retrieval (TBIR), relevant images are to be reranked using visual features after the initial text-based search. In this paper, we propose a new bag-based reranking framework for large-scale TBIR. Specifically, we first cluster relevant images using both textual and visual features. By treating each cluster as a "bag" and the images in the bag as "instances," we formulate this problem as a multi-instance (MI) learning problem. MI learning methods such as mi-SVM can be readily incorporated into our bag-based reranking framework. Observing that at least a certain portion of a positive bag is of positive instances while a negative bag might also contain positive instances, we further use a more suitable generalized MI (GMI) setting for this application. To address the ambiguities on the instance labels in the positive and negative bags under this GMI setting, we develop a new method referred to as GMI-SVM to enhance retrieval performance by propagating the labels from the bag level to the instance level. To acquire bag annotations for (G)MI learning, we propose a bag ranking method to rank all the bags according to the defined bag ranking score. The top ranked bags are used as pseudopositive training bags, while pseudonegative training bags can be obtained by randomly sampling a few irrelevant images that are not associated with the textual query. Comprehensive experiments on the challenging real-world data set NUS-WIDE demonstrate our framework with automatic bag annotation can achieve the best performances compared with existing image reranking methods. Our experiments also demonstrate that GMI-SVM can achieve better performances when using the manually labeled training bags obtained from relevance feedback.

Spectral clustering (SC) methods have been successfully applied to many real-world applications. The success of these SC methods is largely based on the manifold assumption, namely, that two nearby data points in the high-density region of a low-dimensional data manifold have the same cluster label. However, such an assumption might not always hold on high-dimensional data. When the data do not exhibit a clear low-dimensional manifold structure (e.g., high-dimensional and sparse data), the clustering performance of SC will be degraded and become even worse than K -means clustering. In this paper, motivated by the observation that the true cluster assignment matrix for high-dimensional data can be always embedded in a linear space spanned by the data, we propose the spectral embedded clustering (SEC) framework, in which a linearity regularization is explicitly added into the objective function of SC methods. More importantly, the proposed SEC framework can naturally deal with out-of-sample data. We also present a new Laplacian matrix constructed from a local regression of each pattern and incorporate it into our SEC framework to capture both local and global discriminative information for clustering. Comprehensive experiments on eight real-world high-dimensional datasets demonstrate the effectiveness and advantages of our SEC framework over existing SC methods and K-means-based clustering methods. Our SEC framework significantly outperforms SC using the Nyström algorithm on unseen data.

No abstract available.

Semisupervised dimensionality reduction has been attracting much attention as it not only utilizes both labeled and unlabeled data simultaneously, but also works well in the situation of out-of-sample. This paper proposes an effective approach of semisupervised dimensionality reduction through label propagation and label regression. Different from previous efforts, the new approach propagates the label information from labeled to unlabeled data with a well-designed mechanism of random walks, in which outliers are effectively detected and the obtained virtual labels of unlabeled data can be well encoded in a weighted regression model. These virtual labels are thereafter regressed with a linear model to calculate the projection matrix for dimensionality reduction. By this means, when the manifold or the clustering assumption of data is satisfied, the labels of labeled data can be correctly propagated to the unlabeled data; and thus, the proposed approach utilizes the labeled and the unlabeled data more effectively than previous work. Experimental results are carried out upon several databases, and the advantage of the new approach is well demonstrated.

Web photos in social media sharing websites such as Flickr are generally accompanied by rich but noisy textual descriptions (tags, captions, categories, etc.). In this paper, we proposed a tag-based photo retrieval framework to improve the retrieval performance for Flickr photos by employing a novel batch mode re-tagging method. The proposed batch mode re-tagging method can automatically refine noisy tags of a group of Flickr photos uploaded by the same user within a short period by leveraging millions of training web images and their associated rich textual descriptions. Specifically, for one group of Flickr photos, we construct a group-specific lexicon which contains only the tags of all photos within the group. For each query tag, we employ the inverted file method to automatically find loosely labeled training web images. We propose a SVM with Augmented Features, referred to as AFSVM, to learn adapted classifiers to refine the annotation tags of photos by leveraging the existing SVM classifiers of popular tags, which are associated with a large amount of positive training web images. Moreover, to further refine the annotation tags of photos in the same group, we additionally introduce an objective function that utilizes the visual similarities of photos within the group as well as the semantic proximities of their tags. Based on the refined tags, photos can be retrieved according to more reliable relevance scores. Extensive experiments demonstrate the effectiveness of our framework.

We propose a visual event recognition framework for consumer domain videos by leveraging a large amount of loosely labeled web videos (e.g., from YouTube). First, we propose a new aligned space-time pyramid matching method to measure the distances between two video clips, where each video clip is divided into space-time volumes over multiple levels. We calculate the pair-wise distances between any two volumes and further integrate the information from different volumes with Integer-flow Earth Mover's Distance (EMD) to explicitly align the volumes. Second, we propose a new cross-domain learning method in order to 1) fuse the information from multiple pyramid levels and features (i.e., space-time feature and static SIFT feature) and 2) cope with the considerable variation in feature distributions between videos from two domains (i.e., web domain and consumer domain). For each pyramid level and each type of local features, we train a set of SVM classifiers based on the combined training set from two domains using multiple base kernels of different kernel types and parameters, which are fused with equal weights to obtain an average classifier. Finally, we propose a cross-domain learning method, referred to as Adaptive Multiple Kernel Learning (A-MKL), to learn an adapted classifier based on multiple base kernels and the prelearned average classifiers by minimizing both the structural risk functional and the mismatch between data distributions from two domains. Extensive experiments demonstrate the effectiveness of our proposed framework that requires only a small number of labeled consumer videos by leveraging web data.

We propose a new dimension reduction algorithm in combination with the Gaussian Mixture Model (GMM) based Patch Distribution Feature for human gait recognition. Instead of representing each average silhouette image as its gray-level feature, we first extract local patch features at every pixel of the average silhouette image and train a GMM to describe the distribution of the patches in each image. A Universal Background Model (UBM) is first trained with local patch features from all gallery images, then every gallery or probe image is represented by the distribution parameters (referred to as Patch Distribution Features (PDF)) of the image-specific GMM adapted from the UBM. To cope with the high dimension of the PDF feature, the Regularized Trace Ratio Discriminant Analysis (RTRDA) is developed to find the most discriminant subspaces for gait recognition. Experiments on USF humanID database show that RTRDA significantly outperforms the existing algorithms and achieves the best recognition results among all the previous works on USF humanID database in terms of average rank-1 recognition rate.

Nowadays, increasingly rich and massive social media data (such as texts, images, audios, videos, blogs, and so on) are being posted to the web, including social networking websites (e.g., MySpace, Facebook), photo and video sharing websites (e.g., Flickr, YouTube), and photo forums (e.g., Photosig.com and Photo.net). Recently, researchers from multidisciplinary areas have proposed to use data-driven approaches for multimedia content understanding by leveraging such unlimited web images and videos as well as their associated rich contextual information (e.g., tag, comments, category, title and metadata). In this three hour tutorial, we plan to introduce the important general concepts and themes of this timely topic. We will also review and summarize the recent multimedia content analysis methods using web-scale social media data as well as present insight into the challenges and future directions in this area. Moreover, we will also show extensive demos on image annotation and retrieval by using rich social media data.

We propose a new distance measure for face recognition and human gait recognition. Each probe image (a face image or an average human silhouette image) is represented as a set of local features uniformly sampled over a grid with fixed spacing, and each gallery image is represented as a set of local features sampled at each pixel. We formulate an integer programming problem to compute the distance (referred to as the image-to-class distance) from one probe image to all the gallery images belonging to a certain class, in which any feature of the probe image can be matched to only one feature from one of the gallery images. Considering computational efficiency as well as the fact that face images or average human silhouette images are roughly aligned in the preprocessing step, we also enforce a spatial neighborhood constraint by only allowing neighboring features that are within a given spatial distance to be considered for feature matching. The integer programming problem is further treated as a classical minimum-weight bipartite graph matching problem, which can be efficiently solved with the Kuhn-Munkres algorithm. We perform comprehensive experiments on three benchmark face databases: 1) the CMU PIE database; 2) the FERET database; and 3) the FRGC database, as well as the USF Human ID gait database. The experiments clearly demonstrate the effectiveness of our image-to-class distance.

A new framework, termed spatially aligned pyramid matching, is proposed for near duplicate image identification. The proposed method robustly handles spatial shifts as well as scale changes, and is extensible for video data. Images are divided into both overlapped and non-overlapped blocks over multiple levels. In the first matching stage, pairwise distances between blocks from the examined image pair are computed using earth mover's distance (EMD) or the visual word with$\chi^{2}$distance based method with scale-invariant feature transform (SIFT) features. In the second stage, multiple alignment hypotheses that consider piecewise spatial shifts and scale variation are postulated and resolved using integer-flow EMD. Moreover, to compute the distances between two videos, we conduct the third step matching (i.e., temporal matching) after spatial matching. Two application scenarios are addressed—near duplicate retrieval (NDR) and near duplicate detection (NDD). For retrieval ranking, a pyramid-based scheme is constructed to fuse matching results from different partition levels. For NDD, we also propose a dual-sample approach by using the multilevel distances as features and support vector machine for binary classification. The proposed methods are shown to clearly outperform existing methods through extensive testing on the Columbia Near Duplicate Image Database and two new datasets. In addition, we also discuss in depth our framework in terms of the extension for video NDR and NDD, the sensitivity to parameters, the utilization of multiscale dense SIFT descriptors, and the test of scalability in image NDD.

In this paper, we propose a new method to recognize gestures of cartoon images with two practical applications, i.e., content-based cartoon image retrieval and interactive cartoon clip synthesis. Upon analyzing the unique properties of four types of features including global color histogram, local color histogram (LCH), edge feature (EF), and motion direction feature (MDF), we propose to employ different features for different purposes and in various phases. We use EF to define a graph and then refine its local structure by LCH. Based on this graph, we adopt a transductive learning algorithm to construct local patches for each cartoon image. A spectral method is then proposed to optimize the local structure of each patch and then align these patches globally. MDF is fused with EF and LCH and a cartoon gesture space is constructed for cartoon image gesture recognition. We apply the proposed method to content-based cartoon image retrieval and interactive cartoon clip synthesis. The experiments demonstrate the effectiveness of our method.

We propose a unified manifold learning framework for semi-supervised and unsupervised dimension reduction by employing a simple but effective linear regression function to map the new data points. For semi-supervised dimension reduction, we aim to find the optimal prediction labels F for all the training samples X, the linear regression function h(X) and the regression residue F(0) = F - h(X) simultaneously. Our new objective function integrates two terms related to label fitness and manifold smoothness as well as a flexible penalty term defined on the residue F(0). Our Semi-Supervised learning framework, referred to as flexible manifold embedding (FME), can effectively utilize label information from labeled data as well as a manifold structure from both labeled and unlabeled data. By modeling the mismatch between h(X) and F, we show that FME relaxes the hard linear constraint F = h(X) in manifold regularization (MR), making it better cope with the data sampled from a nonlinear manifold. In addition, we propose a simplified version (referred to as FME/U) for unsupervised dimension reduction. We also show that our proposed framework provides a unified view to explain and understand many semi-supervised, supervised and unsupervised dimension reduction techniques. Comprehensive experiments on several benchmark databases demonstrate the significant improvement over existing dimension reduction algorithms.

In this paper, we propose a new image clustering algorithm, referred to as clustering using local discriminant models and global integration (LDMGI). To deal with the data points sampled from a nonlinear manifold, for each data point, we construct a local clique comprising this data point and its neighboring data points. Inspired by the Fisher criterion, we use a local discriminant model for each local clique to evaluate the clustering performance of samples within the local clique. To obtain the clustering result, we further propose a unified objective function to globally integrate the local models of all the local cliques. With the unified objective function, spectral relaxation and spectral rotation are used to obtain the binary cluster indicator matrix for all the samples. We show that LDMGI shares a similar objective function with the spectral clustering (SC) algorithms, e.g., normalized cut (NCut). In contrast to NCut in which the Laplacian matrix is directly calculated based upon a Gaussian function, a new Laplacian matrix is learnt in LDMGI by exploiting both manifold structure and local discriminant information. We also prove that K-means and discriminative K-means (DisKmeans) are both special cases of LDMGI. Extensive experiments on several benchmark image datasets demonstrate the effectiveness of LDMGI. We observe in the experiments that LDMGI is more robust to algorithmic parameter, when compared with NCut. Thus, LDMGI is more appealing for the real image clustering applications in which the ground truth is generally not available for tuning algorithmic parameters.

Cross-domain learning methods have shown promising results by leveraging labeled patterns from auxiliary domains to learn a robust classifier for target domain, which has a limited number of labeled samples. To cope with the tremendous change of feature distribution between different domains in video concept detection, we propose a new cross-domain kernel learning method. Our method, referred to as Domain Transfer SVM (DTSVM), simultaneously learns a kernel function and a robust SVM classifier by minimizing both the structural risk functional of SVM and the distribution mismatch of labeled and unlabeled samples between the auxiliary and target domains. Comprehensive experiments on the challenging TRECVID corpus demonstrate that DTSVM outperforms existing cross-domain learning and multiple kernel learning methods.

We propose a multiple source domain adaptation method, referred to as Domain Adaptation Machine (DAM), to learn a robust decision function (referred to as target classifier) for label prediction of patterns from the target domain by leveraging a set of pre-computed classifiers (referred to as auxiliary/source classifiers) independently learned with the labeled patterns from multiple source domains. We introduce a new data-dependent regularizer based on smoothness assumption into Least-Squares SVM (LS-SVM), which enforces that the target classifier shares similar decision values with the auxiliary classifiers from relevant source domains on the unlabeled patterns of the target domain. In addition, we employ a sparsity regularizer to learn a sparse target classifier. Comprehensive experiments on the challenging TRECVID 2005 corpus demonstrate that DAM outperforms the existing multiple source domain adaptation methods for video concept detection in terms of effectiveness and efficiency.

The ACM SIGMM International Workshop on Social Media (WSM'09) is the first workshop held in conjunction with the ACM International Multimedia Conference (MM'09) at Bejing, P.R. China, 2009. This workshop provides a forum for researchers and practitioners from all over the world to share information on their latest investigations on social media analysis, exploration, search, mining, and emerging new social media applications.

The rapid popularization of digital cameras and mobile phone cameras has lead to an explosive growth of consumer photo collections. In this paper, we present a (quasi) real-time textual query based personal photo retrieval system by leveraging millions of web images and their associated rich textual descriptions (captions, categories, etc.). After a user provides a textual query (e.g., "pool"), our system exploits the inverted file method to automatically find the positive web images that are related to the textual query "pool" as well as the negative web images which are irrelevant to the textual query. Based on these automatically retrieved relevant and irrelevant web images, we employ two simple but effective classification methods, k Nearest Neighbor (kNN) and decision stumps, to rank personal consumer photos. To further improve the photo retrieval performance, we propose three new relevance feedback methods via cross-domain learning. These methods effectively utilize both the web images and the consumer images. In particular, our proposed cross-domain learning methods can learn robust classifiers with only a very limited amount of labeled consumer photos from the user by leveraging the pre-learned decision stumps at interactive response time. Extensive experiments on both consumer and professional stock photo datasets demonstrated the effectiveness and efficiency of our system, which is also inherently not limited by any predefined lexicon.

In this demonstration, we present a (quasi) real-time textual query based image retrieval system (T-IRS) for consumer photos by leveraging millions of web images and their associated rich textual descriptions (captions, categories, etc.). After a user provides a textual query (e.g., boat), our system automatically finds the positive web images that are related to the textual query boat as well as the negative web images which are irrelevant to the textual query. Based on these automatically retrieved positive and negative web images, we employ the decision stump ensemble classifier to rank personal consumer photos. To further improve the photo retrieval performance, we also develop a novel relevance feedback method, referred to as Cross-Domain Regularized Regression (CDRR), which effectively utilizes both the web images and the consumer images. Our system is inherently not limited by any predefined lexicon.

Rich multimedia content including images, audio and text are frequently used to describe the same semantics in E-Learning and Ebusiness web pages, instructive slides, multimedia cyclopedias, and so on. In this paper, we present a framework for cross-media retrieval, where the query example and the retrieved result(s) can be of different media types. We first construct Multimedia Correlation Space (MMCS) by exploring the semantic correlation of different multimedia modalities, during which multimedia content and co-occurrence information is utilized. We propose a novel ranking algorithm, namely ranking with Local Regression and Global Alignment (LRGA), which learns a robust Laplacian matrix for data ranking. In LRGA, for each data point, a local linear regression model is used to predict the ranking values of its neighboring points. We propose a unified objective function to globally align the local models from all the data points so that an optimal ranking value can be assigned to each data point. LRGA is insensitive to parameters, making it particularly suitable for data ranking. A relevance feedback algorithm is proposed to improve the retrieval performance. Comprehensive experiments have demonstrated the effectiveness of our methods.

Cartoons play important roles in many areas, but it requires a lot of labor to produce new cartoon clips. In this paper, we propose a gesture recognition method for cartoon character images with two applications, namely content-based cartoon image retrieval and cartoon clip synthesis. We first define Edge Features (EF) and Motion Direction Features (MDF) for cartoon character images. The features are classified into two different groups, namely intra-features and inter-features. An Unsupervised Bi-Distance Metric Learning (UBDML) algorithm is proposed to recognize the gestures of cartoon character images. Different from the previous research efforts on distance metric learning, UBDML learns the optimal distance metric from the heterogeneous distance metrics derived from intra-features and inter-features. Content-based cartoon character image retrieval and cartoon clip synthesis can be carried out based on the distance metric learned by UBDML. Experiments show that the cartoon character image retrieval has a high precision and that the cartoon clip synthesis can be carried out efficiently.

The success of bilinear subspace learning heavily depends on reducing correlations among features along rows and columns of the data matrices. In this work, we study the problem of rearranging elements within a matrix in order to maximize these correlations so that information redundancy in matrix data can be more extensively removed by existing bilinear subspace learning algorithms. An efficient iterative algorithm is proposed to tackle this essentially integer programming problem. In each step, the matrix structure is refined with a constrained Earth Mover's Distance procedure that incrementally rearranges matrices to become more similar to their low-rank approximations, which have high correlation among features along rows and columns. In addition, we present two extensions of the algorithm for conducting supervised bilinear subspace learning. Experiments in both unsupervised and supervised bilinear subspace learning demonstrate the effectiveness of our proposed algorithms in improving data compression performance and classification accuracy.

Recent research has demonstrated the success of tensor based subspace learning in both unsupervised and supervised configurations (e.g., 2-D PCA, 2-D LDA, and DATER). In this correspondence, we present a new semi-supervised subspace learning algorithm by integrating the tensor representation and the complementary information conveyed by unlabeled data. Conventional semi-supervised algorithms mostly impose a regularization term based on the data representation in the original feature space. Instead, we utilize graph Laplacian regularization based on the low-dimensional feature space. An iterative algorithm, referred to as adaptive regularization based semi-supervised discriminant analysis with tensor representation (ARSDA/T), is also developed to compute the solution. In addition to handling tensor data, a vector-based variant (ARSDA/V) is also presented, in which the tensor data are converted into vectors before subspace learning. Comprehensive experiments on the CMU PIE and YALE-B databases demonstrate that ARSDA/T brings significant improvement in face recognition accuracy over both conventional supervised and semi-supervised subspace learning algorithms.

A new framework, termed spatially aligned pyramid matching, is proposed for near duplicate image identification. The proposed method robustly handles spatial shifts as well as scale changes. Images are divided into both overlapped and non-overlapped blocks over multiple levels. In the first matching stage, pairwise distances between blocks from the examined image pair are computed using SIFT features and Earth Moverpsilas distance (EMD). In the second stage, multiple alignment hypotheses that consider piecewise spatial shifts and scale variation are postulated and resolved using integer-flow EMD. Two application scenarios are addressed - retrieval ranking and binary classification. For retrieval ranking, a pyramid-based scheme is constructed to fuse matching results from different partition levels. For binary classification, a novel generalized neighborhood component analysis method is formulated that can be effectively used in tandem with SVMs to select the most critical matching components. The proposed methods are shown to clearly outperform existing methods through extensive testing on the Columbia near duplicate image database and another new dataset.

In this work, we systematically study the problem of event recognition in unconstrained news video sequences. We adopt the discriminative kernel-based method for which video clip similarity plays an important role. First, we represent a video clip as a bag of orderless descriptors extracted from all of the constituent frames and apply the earth mover's distance (EMD) to integrate similarities among frames from two clips. Observing that a video clip is usually comprised of multiple subclips corresponding to event evolution over time, we further build a multilevel temporal pyramid. At each pyramid level, we integrate the information from different subclips with Integer-value-constrained EMD to explicitly align the subclips. By fusing the information from the different pyramid levels, we develop Temporally Aligned Pyramid Matching (TAPM) for measuring video similarity. We conduct comprehensive experiments on the TRECVID 2005 corpus, which contains more than 6,800 clips. Our experiments demonstrate that 1) the TAPM multilevel method clearly outperforms single-level EMD (SLEMD) and 2) SLEMD outperforms keyframe and multiframe-based detection methods by a large margin. In addition, we conduct in-depth investigation of various aspects of the proposed techniques such as weight selection in SLEMD, sensitivity to temporal clustering, the effect of temporal alignment, and possible approaches for speed up. Extensive analysis of the results also reveals intuitive interpretation of video event recognition through video subclip alignment at different levels.

I NTEREST from industry and academia has increased dramatically over recent years in the challenging area of event analysis and recognition from various video sources including sports, surveillance, user-generated video, etc. Video event analysis and recognition is a critical task in many applications such as detection of sporting highlights, incident detection in surveillance video, indexing, retrieval and summarization of video databases, and human-computer interaction. This special issue aims to capture the latest advances by the research community working in the area of video event analysis. The call for papers was enthusiastically greeted by the research community and we received over seventy submissions. The special issue presents 16 articles which provide fundamental contributions in a wide range of topics in video event analysis: 1) human action and activity recognition; 2) motion trajectory analysis; 3) video content analysis and pattern mining; 4) audio-visual multi-modal analysis; and 5) video analysis applications. An overview of the organization and a brief summary of the articles selected for publication in the special issue are provided below.

Recent research has demonstrated the success of supervised dimensionality reduction algorithms 2DLDA and 2DMFA, which are based on the image-as-matrix representation, in small sample size cases. To solve the convergence problem in 2DLDA and 2DMFA, we propose in this work two new schemes, called Null Space based 2DLDA (NS2DLDA) and Null Space based 2DMFA (NS2DMFA), and apply them to the challenging multi-view face recognition task. First, we convert each 2-D face image (matrix) into a vector and compute the first projection matrixP1from the null space of the intra-class scatter matrix, such that the samples from the same class are projected to the same point. Then the data are projected and reconstructed withP1. Finally, we re-organize the reconstructed datum into a matrix and then compute the second projection directionP2, in the form of a Kronecker product of two matrices, by maximizing the inter-class scatter. A proof of algorithmic convergence is provided. The experiments on two benchmark multi-view face databases, the CMU PIE and FERET databases, demonstrate that NS2DLDA outperforms Fisherface, Null Space LDA (NSLDA) and 2DLDA. Additionally, NS2DMFA is also demonstrated to be more accurate than MFA and 2DMFA for face recognition.

Principal Components Analysis (PCA) has traditionally been utilized with data expressed in the form of 1-D vectors, but there exists much data such as gray-level images, video sequences, Gabor-filtered images and so on, that are intrinsically in the form of second or higher order tensors. For representations of image objects in their intrinsic form and order rather than concatenating all the object data into a single vector, we propose in this paper a new optimal object reconstruction criterion with which the information of a high-dimensional tensor is represented as a much lower dimensional tensor computed from projections to multiple concurrent subspaces. In each of these subspaces, correlations with respect to one of the tensor dimensions are reduced, enabling better object reconstruction performance. Concurrent subspaces analysis (CSA) is presented to efficiently learn these subspaces in an iterative manner. In contrast to techniques such as PCA which vectorize tensor data, CSA's direct use of data in tensor form brings an enhanced ability to learn a representative subspace and an increased number of available projection directions. These properties enable CSA to outperform traditional algorithms in the common case of small sample sizes, where CSA can be effective even with only a single sample per class. Extensive experiments on images of faces and digital numbers encoded as second or third order tensors demonstrate that the proposed CSA outperforms PCA-based algorithms in object reconstruction and object recognition.

Face recognition is a challenging problem, especially when the face images are not strictly aligned (e.g., images can be captured from different viewpoints or the faces may not be accurately cropped by a human or automatic algorithm). In this correspondence, we investigate face recognition under the scenarios with potential spatial misalignments. First, we formulate an asymmetric similarity measure based on Spatially constrained Earth Mover's Distance (SEMD), for which the source image is partitioned into nonoverlapping local patches while the destination image is represented as a set of overlapping local patches at different positions. Assuming that faces are already roughly aligned according to the positions of their eyes, one patch in the source image can be matched only to one of its neighboring patches in the destination image under the spatial constraint of reasonably small misalignments. Because the similarity measure as defined by SEMD is asymmetric, we propose two schemes to combine the two similarity measures computed in both directions. Moreover, we adopt a distance-as-feature approach by treating the distances to the reference images as features in a Kernel Discriminant Analysis (KDA) framework. Experiments on three benchmark face databases, namely the CMU PIE, FERET, and FRGC databases, demonstrate the effectiveness of the proposed SEMD.

Graph-embedding along with its linearization and kernelization provides a general framework that unifies most traditional dimensionality reduction algorithms. From this framework, we propose a new manifold learning technique called discriminant locally linear embedding (DLLE), in which the local geometric properties within each class are preserved according to the locally linear embedding (LLE) criterion, and the separability between different classes is enforced by maximizing margins between point pairs on different classes. To deal with the out-of-sample problem in visual recognition with vector input, the linear version of DLLE, i.e., linearization of DLLE (DLLE/L), is directly proposed through the graph-embedding framework. Moreover, we propose its multilinear version, i.e., tensorization of DLLE, for the out-of-sample problem with high-order tensor input. Based on DLLE, a procedure for gait recognition is described. We conduct comprehensive experiments on both gait and face recognition, and observe that: 1) DLLE along its linearization and tensorization outperforms the related versions of linear discriminant analysis, and DLLE/L demonstrates greater effectiveness than the linearization of LLE; 2) algorithms based on tensor representations are generally superior to linear algorithms when dealing with intrinsically high-order data; and 3) for human gait recognition, DLLE/L generally obtains higher accuracy than state-of-the-art gait recognition algorithms on the standard University of South Florida gait database.

Human gait is a promising biometrics resource. In this paper, the information about gait is obtained from the motions of the different parts of the silhouette. The human silhouette is segmented into seven components, namely head, arm, trunk, thigh, front-leg, back-leg, and feet. The leg silhouettes for the front-leg and the back-leg are considered separately because, during walking, the left leg and the right leg are in front or at the back by turns. Each of the seven components and a number of combinations of the components are then studied with regard to two useful applications: human identification (ID) recognition and gender recognition. More than 500 different experiments on human ID and gender recognition are carried out under a wide range of circumstances. The effectiveness of the seven human gait components for ID and gender recognition is analyzed.

A large family of algorithms for dimensionality reduction end with solving a Trace Ratio problem in the form of arg maxWTr(WTSPW)/Tr(WT SIW)1, which is generally transformed into the corresponding Ratio Trace form arg maxWTr[ (WTSIW)-1(WTSPW) ] for obtaining a closed-form but inexact solution. In this work, an efficient iterative procedure is presented to directly solve the Trace Ratio problem. In each step, a Trace Difference problem arg maxWTr [WT(SP- lambdaSI) W] is solved with lambda being the trace ratio value computed from the previous step. Convergence of the projection matrix W, as well as the global optimum of the trace ratio value lambda, are proven based on point-to-set map theories. In addition, this procedure is further extended for solving trace ratio problems with more general constraint WTCW=I and providing exact solutions for kernel-based subspace learning problems. Extensive experiments on faces and UCI data demonstrate the high convergence speed of the proposed solution, as well as its superiority in classification capability over corresponding solutions to the ratio trace problem.

In this work, we systematically study the problem of visual event recognition in unconstrained news video sequences. We adopt the discriminative kernel-based method for which video clip similarity plays an important role. First, we represent a video clip as a bag of orderless descriptors extracted from all of the constituent frames and apply Earth mover's distance (EMD) to integrate similarities among frames from two clips. Observing that a video clip is usually comprised of multiple sub-clips corresponding to event evolution over time, we further build a multilevel temporal pyramid. At each pyramid level, we integrate the information from different sub-clips with Integer-value-constrained EMD to explicitly align the sub-clips. By fusing the information from the different pyramid levels, we develop temporally aligned pyramid matching (TAPM) for measuring video similarity. We conduct comprehensive experiments on the Trecvid 2005 corpus, which contains more than 6,800 clips. Our experiments demonstrate that 1) the TAPM multi-level method clearly outperforms single-level EMD, and 2) single-level EMD outperforms by a large margin (43.0% in Mean Average Precision) basic detection methods that use only a single key-frame. Extensive analysis of the results also reveals an intuitive interpretation of subclip alignment at different levels.

The success of tensor-based subspace learning depends heavily on reducing correlations along the column vectors of the mode-k flattened matrix. In this work, we study the problem of rearranging elements within a tensor in order to maximize these correlations, so that information redundancy in tensor data can be more extensively removed by existing tensor-based dimensionality reduction algorithms. An efficient iterative algorithm is proposed to tackle this essentially integer optimization problem. In each step, the tensor structure is refined with a spatially-constrained Earth Mover's Distance procedure that incrementally rearranges tensors to become more similar to their low rank approximations, which have high correlation among features along certain tensor dimensions. Monotonic convergence of the algorithm is proven using an auxiliary function analogous to that used for proving convergence of the Expectation-Maximization algorithm. In addition, we present an extension of the algorithm for conducting supervised subspace learning with tensor data. Experiments in both unsupervised and supervised subspace learning demonstrate the effectiveness of our proposed algorithms in improving data compression performance and classification accuracy.

A large family of algorithms - supervised or unsupervised; stemming from statistics or geometry theory - has been designed to provide different solutions to the problem of dimensionality reduction. Despite the different motivations of these algorithms, we present in this paper a general formulation known as graph embedding to unify them within a common framework. In graph embedding, each algorithm can be considered as the direct graph embedding or its linear/kernel/tensor extension of a specific intrinsic graph that describes certain desired statistical or geometric properties of a data set, with constraints from scale normalization or a penalty graph that characterizes a statistical or geometric property that should be avoided. Furthermore, the graph embedding framework can be used as a general platform for developing new dimensionality reduction algorithms. By utilizing this framework as a tool, we propose a new supervised dimensionality reduction algorithm called marginal Fisher analysis in which the intrinsic graph characterizes the intraclass compactness and connects each data point with its neighboring points of the same class, while the penalty graph connects the marginal points and characterizes the interclass separability. We show that MFA effectively overcomes the limitations of the traditional linear discriminant analysis algorithm due to data distribution assumptions and available projection directions. Real face recognition experiments show the superiority of our proposed MFA in comparison to LDA, also for corresponding kernel and tensor extensions

Traditional manifold learning algorithms, such as ISOMAP, LLE, and Laplacian Eigenmap, mainly focus on uncovering the latent low-dimensional geometry structure of the training samples in an unsupervised manner where useful class information is ignored. Therefore, the derived low-dimensional representations are not necessarily optimal in discriminative capability. In this paper, we study the discriminant analysis problem by considering the nonlinear manifold structure of data space. To this end, firstly, a new clustering algorithm, called Intra-Cluster Balanced K-Means (ICBKM), is proposed to partition the samples into multiple clusters while ensure that there are balanced samples for the classes within each cluster; approximately, each cluster can be considered as a local patch on the embedded manifold. Then, the local discriminative projections for different clusters are simultaneously calculated by optimizing the global Fisher Criterion based on the cluster weighted data representation. Compared with traditional linear/kernel discriminant analysis (KDA) algorithms, our proposed algorithm has the following characteristics: 1) it essentially is a KDA algorithm with specific geometry-adaptive-kernel tailored to the specific data structure, in contrast to traditional KDA in which the kernel is fixed and independent to the data set; 2) it is approximately a locally linear while globally nonlinear discriminant analyzer; 3) it does not need to store the original samples for computing the low-dimensional representation of a new data; and 4) it is computationally efficient compared with traditional KDA when the sample number is large. The toy problem on artificial data demonstrates the effectiveness of our proposed algorithm in deriving discriminative representations for problems with nonlinear classification hyperplane. The face recognition experiments on YALE and CMU PIE databases show that our proposed algorithm significantly outperforms linear discriminant analysis (LDA) as well as Mixture LDA, and has higher accuracy than KDA with traditional kernels

Dimensionality reduction algorithms, which aim to select a small set of efficient and discriminant features, have attracted great attention for human gait recognition and content-based image retrieval (CBIR). In this paper, we present extensions of our recently proposed marginal Fisher analysis (MFA) to address these problems. For human gait recognition, we first present a direct application of MFA, then inspired by recent advances in matrix and tensor-based dimensionality reduction algorithms, we present matrix-based MFA for directly handling 2-D input in the form of gray-level averaged images. For CBIR, we deal with the relevance feedback problem by extending MFA to marginal biased analysis, in which within-class compactness is characterized only by the distances between each positive sample and its neighboring positive samples. In addition, we present a new technique to acquire a direct optimal solution for MFA without resorting to objective function modification as done in many previous algorithms. We conduct comprehensive experiments on the USF HumanID gait database and the Corel image retrieval database. Experimental results demonstrate that MFA and its extensions outperform related algorithms in both applications.

The process of face verification is guided by a pre-learned global threshold, which, however, is often inconsistent with class-specific optimal thresholds. It is, hence, beneficial to pursue a balance of the class-specific thresholds in the model-learning stage. In this paper, we present a new dimensionality reduction algorithm tailored to the verification task that ensures threshold balance. This is achieved by the following aspects. First, feasibility is guaranteed by employing an affine transformation matrix, instead of the conventional projection matrix, for dimensionality reduction, and, hence, we call the proposed algorithm threshold balanced transformation (TBT). Then, the affine transformation matrix, constrained as the product of an orthogonal matrix and a diagonal matrix, is optimized to improve the threshold balance and classification capability in an iterative manner. Unlike most algorithms for face verification which are directly transplanted from face identification literature, TBT is specifically designed for face verification and clarifies the intrinsic distinction between these two tasks. Experiments on three benchmark face databases demonstrate that TBT significantly outperforms the state-of-the-art subspace techniques for face verification.

There is a growing interest in subspace learning techniques for face recognition; however, the excessive dimension of the data space often brings the algorithms into the curse of dimensionality dilemma. In this paper, we present a novel approach to solve the supervised dimensionality reduction problem by encoding an image object as a general tensor of second or even higher order. First, we propose a discriminant tensor criterion, whereby multiple interrelated lower dimensional discriminative subspaces are derived for feature extraction. Then, a novel approach, called k-mode optimization, is presented to iteratively learn these subspaces by unfolding the tensor along different tensor directions. We call this algorithm multilinear discriminant analysis (MDA), which has the following characteristics: 1) multiple interrelated subspaces can collaborate to discriminate different classes, 2) for classification problems involving higher order tensors, the MDA algorithm can avoid the curse of dimensionality dilemma and alleviate the small sample size problem, and 3) the computational cost in the learning stage is reduced to a large extent owing to the reduced data dimensions in k-mode optimization. We provide extensive experiments on ORL, CMU PIE, and FERET databases by encoding face images as second- or third-order tensors to demonstrate that the proposed MDA algorithm based on higher order tensors has the potential to outperform the traditional vector-based subspace learning algorithms, especially in the cases with small sample sizes.

In supervised dimensionality reduction, tensor representations of images have recently been employed to enhance classification of high dimensional data with small training sets. Previous approaches for handling tensor data have been formulated with tight restrictions on projection directions that, along with convergence issues and the assumption of Gaussian-distributed class data, limit its face-recognition performance. To overcome these problems, we propose a method of rank-one projections with adaptive margins (RPAM) that gives a provably convergent solution for tensor data over a more general class of projections, while accounting for margins between samples of different classes. In contrast to previous margin-based works which determine margin sample pairs within the original high dimensional feature space, RPAM aims instead to maximize the margins defined in the expected lower dimensional feature sub-space by progressive margin refinement after each rank-one projection. In addition to handling tensor data, vector-based variants of RPAM are presented for linear mappings and for nonlinear mappings using kernel tricks. Comprehensive experimental results demonstrate that RPAM brings significant improvement in face recognition over previous subspace learning techniques.

In supervised dimensionality reduction, tensor representations of images have recently been employed to enhance classification of high-dimensional data with small training sets. To handle tensor data, this approach has been formulated with tight restrictions on projection directions that, along with convergence issues and the assumption of Gaussian distributed class data, limits its face recognition performance. To overcome these problems, we propose a method of rank-one projections with adaptive margins (RPAM) that gives a provably convergent solution for tensor data over a more general class of projections, while accounting for margins between samples of different classes. In contrast to previous margin based works which determine margin sample pairs within the original high dimensional space, RPAM instead aims to maximize the margins defined in the expected lower dimensional feature subspace by progressive margin refinement after each rank-one projection. In addition to handling tensor data, vector-based variants of RPAM are presented for linear mappings and for nonlinear mappings using kernel tricks. Comprehensive experimental results demonstrate that RPAM brings significant improvement in face recognition over previous subspace learning techniques.

Human gait is an important biometric feature. It can be perceived from a great distance and has recently attracted greater attention in video-surveillance-related applications, such as closed-circuit television. We explore gait recognition based on a matrix representation in this paper. First, binary silhouettes over one gait cycle are averaged. As a result, each gait video sequence, containing a number of gait cycles, is represented by a series of gray-level averaged images. Then, a matrix-based unsupervised algorithm, namely coupled subspace analysis (CSA), is employed as a preprocessing step to remove noise and retain the most representative information. Finally, a supervised algorithm, namely discriminant analysis with tensor representation, is applied to further improve classification ability. This matrix-based scheme demonstrates a much better gait recognition performance than state-of-the-art algorithms on the standard USF HumanID Gait database.

A representative subspace is significant for image analysis, while the corresponding techniques often suffer from the curse of dimensionality dilemma. In this paper, we propose a new algorithm, called concurrent subspaces analysis (CSA), to derive representative subspaces by encoding image objects as 2/sup nd/ or even higher order tensors. In CSA, an original higher dimensional tensor is transformed into a lower dimensional one using multiple concurrent subspaces that characterize the most representative information of different dimensions, respectively. Moreover, an efficient procedure is provided to learn these subspaces in an iterative manner. As analyzed in this paper, each sub-step of CSA takes the column vectors of the matrices, which are acquired from the k-mode unfolding of the tensors, as the new objects to be analyzed, thus the curse of dimensionality dilemma can be effectively avoided. The extensive experiments on the 3/sup rd/ order tensor data, simulated video sequences and Gabor filtered digital number image database show that CSA outperforms principal component analysis in terms of both reconstruction and classification capability.

In this paper, we present a novel approach to solving the supervised dimensionality reduction problem by encoding an image object as a general tensor of 2nd or higher order. First, we propose a discriminant tensor criterion (DTC), whereby multiple interrelated lower-dimensional discriminative subspaces are derived for feature selection. Then, a novel approach called k-mode cluster-based discriminant analysis is presented to iteratively learn these subspaces by unfolding the tensor along different tensor dimensions. We call this algorithm discriminant analysis with tensor representation (DATER), which has the following characteristics: 1) multiple interrelated subspaces can collaborate to discriminate different classes; 2) for classification problems involving higher-order tensors, the DATER algorithm can avoid the curse of dimensionality dilemma and overcome the small sample size problem; and 3) the computational cost in the learning stage is reduced to a large extent owing to the reduced data dimensions in generalized eigenvalue decomposition. We provide extensive experiments by encoding face images as 2nd or 3rd order tensors to demonstrate that the proposed DATER algorithm based on higher order tensors has the potential to outperform the traditional subspace learning algorithms, especially in the small sample size cases.

In the last decades, a large family of algorithms - supervised or unsupervised; stemming from statistic or geometry theory - have been proposed to provide different solutions to the problem of dimensionality reduction. In this paper, beyond the different motivations of these algorithms, we propose a general framework, graph embedding along with its linearization and kernelization, which in theory reveals the underlying objective shared by most previous algorithms. It presents a unified perspective to understand these algorithms; that is, each algorithm can be considered as the direct graph embedding or its linear/kernel extension of some specific graph characterizing certain statistic or geometry property of a data set. Furthermore, this framework is a general platform to develop new algorithm for dimensionality reduction. To this end, we propose a new supervised algorithm, Marginal Fisher Analysis (MFA), for dimensionality reduction by designing two graphs that characterize the intra-class compactness and inter-class separability, respectively. MFA measures the intra-class compactness with the distance between each data point and its neighboring points of the same class, and measures the inter-class separability with the class margins; thus it overcomes the limitations of traditional Linear Discriminant Analysis algorithm in terms of data distribution assumptions and available projection directions. The toy problem on artificial data and the real face recognition experiments both show the superiority of our proposed MFA in comparison to LDA.

It was prescriptive that an image matrix was transformed into a vector before the kernel-based subspace learning. In this paper, we take the kernel discriminant analysis (KDA) algorithm as an example to perform kernel analysis on 2D image matrices directly. First, each image matrix is decomposed as the product of two orthogonal matrices and a diagonal one by using singular value decomposition; then an image matrix is expanded to be of higher or even infinite dimensions by applying the kernel trick on the column vectors of the two orthogonal matrices; finally, two coupled discriminative kernel subspaces are iteratively learned for dimensionality reduction by optimizing the Fisher criterion measured by Frobenius norm. The derived algorithm, called coupled kernel discriminant analysis (CKDA), effectively utilizes the underlying spatial structure of objects and the discriminating information is encoded in two coupled kernel subspaces respectively. The experiments on real face databases compared with KDA and Fisherface validate the effectiveness of CKDA.

We simultaneously approach two tasks of nonlinear discriminant analysis and kernel selection problem by proposing a unified criterion, Fisher+Kernel criterion. In addition, an efficient procedure is derived to optimize this new criterion in an iterative manner. More specifically, original input vector is first transformed into a higher dimensional feature matrix through a battery of nonlinear mappings involved in different kernels. Then, based on the feature matrices, FKC is presented within two coupled projection spaces: one projection space is used to search for the optimal combinations of kernels; while the other encodes the optimal nonlinear discriminating projection directions. Our proposed method is a unified framework for both kernel selection and nonlinear discriminant analysis. Besides, the algorithm potentially alleviates overfitting problem existing in traditional KDA and has no singularity problems in most cases. The effectiveness of our proposed algorithm is validated by extensive face recognition experiments on several datasets.

The canonical face recognition algorithm Eigenface and Fisherface are both based on one dimensional vector representation. However, with the high feature dimensions and the small training data, face recognition often suffers from the curse of dimension and the small sample problem. Recent research [4] shows that face recognition based on direct 2D matrix representation, i.e. 2DPCA, obtains better performance than that based on traditional vector representation. However, there are three questions left unresolved in the 2DPCA algorithm: I ) what is the meaning of the eigenvalue and eigenvector of the covariance matrix in 2DPCA; 2) why 2DPCA can outperform Eigenface; and 3) how to reduce the dimension after 2DPCA directly. In this paper, we analyze 2DPCA in a different view and proof that is 2DPCA actually a "localized" PCA with each row vector of an image as object. With this explanation, we discover the intrinsic reason that 2DPCA can outperform Eigenface is because fewer feature dimensions and more samples are used in 2DPCA when compared with Eigenface. To further reduce the dimension after 2DPCA, a two-stage strategy, namely parallel image matrix compression (PIMC), is proposed to compress the image matrix redundancy, which exists among row vectors and column vectors. The exhaustive experiment results demonstrate that PIMC is superior to 2DPCA and Eigenface, and PIMC+LDA outperforms 2DPC+LDA and Fisherface.

In this paper, we propose a novel shape interpolation approach based on Poisson equation. We formulate the trajectory problem of shape interpolation as solving Poisson equations defined on a domain mesh. A non-linear gradient field interpolation method is proposed to take both vertex coordinates and surface orientation into account. With proper boundary conditions, the in-between shapes are reconstructed implicitly from the interpolated gradient fields, while traditional methods usually manipulate vertex coordinates directly. Besides of global shape interpolation, our method is also applicable to local shape interpolation, and can be further enhanced by incorporating with deformation. Our approach can generate visual pleasing and physical plausible morphing sequences with stable area and volume changes. Experimental results demonstrate that our technique can avoid the shrinkage problem appeared in linear shape interpolation.

To prevent moving cast shadows from being misunderstood as part of moving objects in change detection based video segmentation, this paper proposes a novel approach to the cast shadow detection based on the edge and region information in multiple frames. First, an initial change detection mask containing moving objects and cast shadows is obtained. Then a Canny edge map is generated. After that, the shadow region is detected and removed through multiframe integration, edge matching, and region growing. Finally, a post processing procedure is used to eliminate noise and tune the boundaries of the objects. Our approach can be used for video segmentation in indoor environment. The experimental results demonstrate its good performance.

To prevent moving cast shadows from being misunderstood as part of moving objects in change detection based video segmentation, this paper proposes a novel approach to cast shadow detection based on the edge and region information in multiple frames. First, an initial change detection mask containing moving objects and cast shadows is obtained. Then, a Canny edge map is generated. After that, the shadow region is detected and removed through multi-frame integration, edge matching and region growing. Finally, a post-processing procedure is used to eliminate noise and tune the boundaries of the objects. Our approach can be used for removing moving cast shadows in indoor environments for better video segmentation. The experimental results demonstrate the good performance of our algorithm.

In this paper, we introduce a novel approach to mesh editing with the Poisson equation as the theoretical foundation. The most distinctive feature of this approach is that it modifies the original mesh geometry implicitly through gradient field manipulation. Our approach can produce desirable and pleasing results for both global and local editing operations, such as deformation, object merging, and smoothing. With the help from a few novel interactive tools, these operations can be performed conveniently with a small amount of user interaction. Our technique has three key components, a basic mesh solver based on the Poisson equation, a gradient field manipulation scheme using local transforms, and a generalized boundary condition representation based on local frames. Experimental results indicate that our framework can outperform previous related mesh editing techniques.