Recent advancements in Large Video Language Models (LVLMs) have highlighted their potential for multi-modal understanding, yet evaluating their factual grounding in videos remains a critical unsolved challenge. To address this gap, we introduce Video SimpleQA, the first comprehensive benchmark tailored for factuality evaluation in video contexts. Our work differs from existing video benchmarks through the following key features: 1) Knowledge required: demanding integration of external knowledge beyond the video’s explicit narrative; 2) Multi-hop fact-seeking question: Each question involves multiple explicit facts and requires strict factual grounding without hypothetical or subjective inferences. We include per-hop single-fact-based sub-QAs alongside final QAs to enable fine-grained, step-by-step evaluation; 3) Short-form definitive answer: Answers are crafted as unambiguous and definitively correct in a short format with minimal scoring variance; 4) Temporal grounded required: Requiring answers to rely on one or more temporal segments in videos, rather than single frames. We extensively evaluate 33 state-of-the-art LVLMs and summarize key findings as follows: 1) Current LVLMs exhibit notable deficiencies in factual adherence, with the best-performing model o3 merely achieving an F-score of 66.3%; 2) Most LVLMs are overconfident in what they generate, with self-stated confidence exceeding actual accuracy; 3) Retrieval-Augmented Generation demonstrates consistent improvements at the cost of additional inference time overhead; 4) Multi-hop QA demonstrates substantially degraded performance compared to single-hop sub-QAs, with first-hop object/event recognition emerging as the primary bottleneck. We position Video SimpleQA as the cornerstone benchmark for video factuality assessment, aiming to steer LVLM development toward verifiable grounding in real-world contexts.

Recent advances in Multi-modal Large Language Models (MLLMs) have showcased remarkable capabilities in vision-language understanding. However, enabling robust video spatial reasoning—the ability to comprehend object locations, orientations, and inter-object relationships in dynamic 3D scenes—remains a key unsolved challenge. Existing approaches primarily rely on spatially grounded supervised fine-tuning or reinforcement learning, yet we observe that such models often exhibit query-locked reasoning, focusing narrowly on objects explicitly mentioned in the prompt while ignoring critical contextual cues. To address this limitation, we propose Object-Centric 3D Rollout (OCR), a novel strategy that introduces structured perturbations to the 3D geometry of selected objects during training. By degrading object-specific visual cues and projecting the altered geometry into 2D space, OCR compels the model to reason holistically across the entire scene. We further design a rollout-based training pipeline that jointly leverages vanilla and region-noisy videos to optimize spatial reasoning trajectories. Experiments demonstrate state-of-the-art performance: our 3B-parameter model achieves 47.5% accuracy on VSI-Bench, outperforming several 7B baselines. Ablations confirm OCR’s superiority over prior rollout strategies (e.g., T-GRPO, NoisyRollout).

Large Language Models (LLMs) demonstrate strong capabilities in broad knowledge representation, yet they are inherently deficient in pixel-level perceptual understanding. Although the Segment Anything Model (SAM) represents a significant advancement in visual-prompt-driven image segmentation, it exhibits notable limitations in multi-mask prediction and category-specific segmentation tasks, and it cannot integrate all segmentation tasks within a unified model architecture. To address these limitations, we present X-SAM, a streamlined Multimodal Large Language Model (MLLM) framework that extends the segmentation paradigm from segment anything to any segmentation. Specifically, we introduce a novel unified framework that enables more advanced pixel-level perceptual comprehension for MLLMs. Furthermore, we propose a new segmentation task, termed Visual GrounDed (VGD) segmentation, which segments all instance objects with interactive visual prompts and empowers MLLMs with visual grounded, pixel-wise interpretative capabilities. To enable effective training on diverse data sources, we present a unified training strategy that supports co-training across multiple datasets. Experimental results demonstrate that X-SAM achieves state-of-the-art performance on a wide range of image segmentation benchmarks, highlighting its efficiency for multimodal, pixel-level visual understanding.

Diffusion-based technologies have made significant strides, particularly in personalized and customized facial generation. However, existing methods struggle to achieve high-fidelity and detailed identity (ID) consistency. This is mainly due to two challenges: insufficient fine-grained control over specific facial areas and the absence of a comprehensive strategy for ID preservation that accounts for both intricate facial details and the overall facial structure. To address these limitations, we introduce ConsistentID, an innovative method crafted for diverse identity-preserving portrait generation under fine-grained multimodal facial prompts, utilizing only a single reference image. ConsistentID comprises two core components: a multimodal facial prompt generator and an ID-preservation network. The facial prompt generator combines localized facial features, facial feature descriptions, and overall facial descriptions to enhance the precision of facial detail reconstruction. The ID-preservation network, optimized with a facial attention localization strategy, ensures consistent identity preservation across facial regions. Together, these components leverage fine-grained multimodal identity information to improve identity preservation accuracy significantly. To drive ConsistentID's training, we propose a fine-grained portrait dataset, FGID, with over 500,000 facial images, offering greater diversity and comprehensiveness than existing public facial datasets. Experimental results substantiate that our ConsistentID achieves exceptional precision and diversity in personalized facial generation, surpassing existing methods in the MyStyle dataset. In addition, although ConsistentID introduces more multimodal ID information, it still maintains rapid inference speed during the generation process.

The rapid advancements in Large Vision Models (LVMs), such as Vision Transformers (ViTs), diffusion models, and visual autoregressive models, have led to an increasing demand for computational resources, resulting in substantial financial and environmental costs. This growing challenge highlights the necessity of developing efficient training methods for LVMs. Progressive learning, a training strategy in which model capacity gradually increases during training, has shown promise in addressing these challenges. In this paper, we take a practical step toward the efficient training of LVMs by automating progressive learning. We focus first on the pre-training of LVMs, using ViTs as a case study. We propose AutoProg-One, an automated progressive learning scheme featuring momentum growth (MoGrow) and the one-shot growth schedule search. Additionally, we extend our approach beyond pre-training to address the transfer learning and fine-tuning of LVMs. We also expand the scope of AutoProg to encompass a wider range of LVMs, including diffusion models and visual autoregressive model. First, we introduce AutoProg-Zero, by enhancing the AutoProg framework with a novel zero-shot automated progressive learning method, eliminating the need for one-shot supernet training. Second, we introduce a novel Unique Stage Identifier (SID) scheme to bridge the gap during network growth. These innovations, integrated with the core principles of AutoProg, offer a comprehensive solution for efficient training across various LVM scenarios. Extensive experiments show that AutoProg accelerates ViT pre-training by up to 1.85 ×on ImageNet and accelerates the fine-tuning of diffusion models, and visual autoregressive model by up to 2.86 × and 1.89 ×, with comparable or even better performance. This work provides a robust and scalable approach to efficient training of LVMs, with potential applications in a wide range of vision tasks.

Recent studies have revealed the potential of training open-source Large Language Models (LLMs) to unleash LLMs' reasoning ability for enhancing vision-language navigation (VLN) performance, and simultaneously mitigate the domain gap between LLMs' training corpus and the VLN task. However, these approaches predominantly adopt straightforward input-output mapping paradigms, causing the mapping learning difficult and the navigational decisions unexplainable. Chain-of-Thought (CoT) training is a promising way to improve both navigational decision accuracy and interpretability, while the complexity of the navigation task makes the perfect CoT labels unavailable and may lead to overfitting through pure CoT supervised fine-tuning. To address these issues, we propose EvolveNav, a novel sElf-improving embodied reasoning paradigm that realizes adaptable and generalizable navigational reasoning for boosting LLM-based vision-language Navigation. Specifically, EvolveNav involves a two-stage training process: (1) Formalized CoT Supervised Fine-Tuning, where we train the model with curated formalized CoT labels to first activate the model's navigational reasoning capabilities, and simultaneously increase the reasoning speed; (2) Self-Reflective Post-Training, where the model is iteratively trained with its own reasoning outputs as self-enriched CoT labels to enhance the supervision diversity. A self-reflective auxiliary task is also designed to encourage the model to learn correct reasoning patterns by contrasting with wrong ones. Experimental results under both task-specific and cross-task training paradigms demonstrate the consistent superiority of EvolveNav over previous LLM-based VLN approaches on various popular benchmarks, including R2R, REVERIE, CVDN, and SOON. EvolveNav open avenues for exploring effective self-improving reasoning paradigms, enabling building agents capable of self-evolving for promoting LLM-based embodied AI research.

In this paper, we address the challenging task of multimodal reasoning by incorporating the notion of "slow thinking" into multimodal large language models (MLLMs). Our core idea is that models can learn to adaptively use different levels of reasoning to tackle questions of varying complexity. We propose a novel paradigm of Self-structured Chain of Thought (SCoT), which consists of minimal semantic atomic steps. Unlike existing methods that rely on structured templates or free-form paradigms, our method not only generates flexible CoT structures for various complex tasks but also mitigates the phenomenon of overthinking for easier tasks. To introduce structured reasoning into visual cognition, we design a novel AtomThink framework with four key modules: (i) a data engine to generate high-quality multimodal reasoning paths; (ii) a supervised fine-tuning (SFT) process with serialized inference data; (iii) a policy-guided multi-turn inference method; and (iv) an atomic capability metric to evaluate the single-step utilization rate. Extensive experiments demonstrate that the proposed AtomThink significantly improves the performance of baseline MLLMs, achieving more than 10% average accuracy gains on MathVista and MathVerse. Compared to state-of-the-art structured CoT approaches, our method not only achieves higher accuracy but also improves data utilization by 5 × and boosts inference efficiency by 85.3%.

Data scarcity is a long-standing challenge in the vision-language navigation (VLN) field, which extremely hinders the generalization of agents to unseen environments. Previous works primarily rely on additional simulator data or web-collected images/videos to improve the generalization. However, the simulator environments still face limited diversity, and the web-collected data often require extensive labor to remove the noise. In this article, we propose a Rewriting-driven AugMentation (RAM) paradigm for VLN, which directly creates the unseen observation-instruction pairs via rewriting human-annotated training data. Benefiting from our rewriting mechanism, new observation-instruction pairs can be obtained in both simulator-free and labor-saving manners to promote generalization. Specifically, we first introduce object-enriched observation rewriting, where we combine vision-language models (VLMs) and large language models (LLMs) to derive rewritten object-enriched scene descriptions, enabling observation synthesis with diverse objects and spatial layouts via text-to-image generation models (T2IMs). Then, we propose observation-contrast instruction rewriting, which generates observation-aligned rewritten instructions by requiring LLMs to reason the difference between original and new observations. We further develop a mixing-then-focusing training strategy with a random observation cropping scheme, effectively enhancing data distribution diversity while suppressing augmentation data noise during training. Experiments on both the discrete environments (R2R, REVERIE, and R4R datasets) and continuous environments (R2R-CE dataset) show the superior performance and impressive generalization ability of our method.

LLM-based agents have demonstrated impressive zero-shot performance in vision-language navigation (VLN) task. However, existing LLM-based methods often focus only on solving high-level task planning by selecting nodes in predefined navigation graphs for movements, overlooking low-level control in navigation scenarios. To bridge this gap, we propose AO-Planner, a novel Affordances-Oriented Planner for continuous VLN task. Our AO-Planner integrates various foundation models to achieve affordances-oriented low-level motion planning and high-level decision-making, both performed in a zero-shot setting. Specifically, we employ a Visual Affordances Prompting (VAP) approach, where the visible ground is segmented by SAM to provide navigational affordances, based on which the LLM selects potential candidate waypoints and plans low-level paths towards selected waypoints. We further propose a high-level PathAgent which marks planned paths into the image input and reasons the most probable path by comprehending all environmental information. Finally, we convert the selected path into 3D coordinates using camera intrinsic parameters and depth information, avoiding challenging 3D predictions for LLMs. Experiments on the challenging R2R-CE and RxR-CE datasets show that AO-Planner achieves state-of-the-art zero-shot performance (8.8% improvement on SPL). Our method can also serve as a data annotator to obtain pseudo-labels, distilling its waypoint prediction ability into a learning-based predictor. This new predictor does not require any waypoint data from the simulator and achieves 47% SR competing with supervised methods. We establish an effective connection between LLM and 3D world, presenting novel prospects for employing foundation models in low-level motion control.

Diffusion models for garment-centric human generation from text or image prompts have garnered emerging attention for their great application potential. However, existing methods often face a dilemma: lightweight approaches, such as adapters, are prone to generate inconsistent textures; while finetune-based methods involve high training costs and struggle to maintain the generalization capabilities of pretrained diffusion models, limiting their performance across diverse scenarios. To address these challenges, we propose DreamFit, which incorporates a lightweight Anything-Dressing Encoder specifically tailored for the garment-centric human generation. DreamFit has three key advantages: (1) Lightweight training: with the proposed adaptive attention and LoRA modules, DreamFit significantly minimizes the model complexity to 83.4M trainable parameters. (2) Anything-Dressing: Our model generalizes surprisingly well to a wide range of (non-)garments, creative styles, and prompt instructions, consistently delivering high-quality results across diverse scenarios. (3) Plug-and-play: DreamFit is engineered for smooth integration with any community control plugins for diffusion models, ensuring easy compatibility and minimizing adoption barriers. To further enhance generation quality, DreamFit leverages pretrained large multi-modal models (LMMs) to enrich the prompt with fine-grained garment descriptions, thereby reducing the prompt gap between training and inference. We conduct comprehensive experiments on both 768 x 512 high-resolution benchmarks and in-the-wild images. DreamFit surpasses all existing methods, highlighting its state-of-the-art capabilities of garment-centric human generation.

Text-Video Retrieval (TVR) aims to align and associate relevant video content with corresponding natural language queries. Most existing TVR methods are based on large-scale pre-trained vision-language models (e.g., CLIP). However, due to CLIP's inherent plain structure, few TVR methods explore the multi-scale representations which offer richer contextual information for a more thorough understanding. To this end, we propose MUSE, a multi-scale mamba with linear computational complexity for efficient cross-resolution modeling. Specifically, the multi-scale representations are generated by applying a feature pyramid on the last single-scale feature map. Then, we employ the Mamba structure as an efficient multi-scale learner to jointly learn scale-wise representations. Furthermore, we conduct comprehensive studies to investigate different model structures and designs. Extensive results on three popular benchmarks have validated the superiority of MUSE.

The rapid development of the autonomous driving industry has led to a significant accumulation of autonomous driving data. Consequently, there comes a growing demand for retrieving data to provide specialized optimization. However, directly applying previous image retrieval methods faces several challenges, such as the lack of global feature representation and inadequate text retrieval ability for complex driving scenes. To address these issues, firstly, we propose the BEV-TSR framework which leverages descriptive text as an input to retrieve corresponding scenes in the Bird’s Eye View (BEV) space. Then to facilitate complex scene retrieval with extensive text descriptions, we employ a large language model (LLM) to extract the semantic features of the text inputs and incorporate knowledge graph embeddings to enhance the semantic richness of the language embedding. To achieve feature alignment between the BEV feature and language embedding, we propose Shared Cross-modal Embedding with a set of shared learnable embeddings to bridge the gap between these two modalities, and employ a caption generation task to further enhance the alignment. Furthermore, there lack of well-formed retrieval datasets for effective evaluation. To this end, we establish a multi-level retrieval dataset, nuScenes-Retrieval, based on the widely adopted nuScenes dataset. Experimental results on the multi-level nuScenes-Retrieval show that BEV-TSR achieves state-of-the-art performance, e.g., 85.78% and 87.66% top-1 accuracy on scene-to-test and text-to-scene retrieval respectively.

Vision-And-Language Navigation (VLN) suffers from the limited diversity and scale of training data, primarily constrained by the manual curation of existing simulators. To address this, we introduce RoomTour3D, a video-instruction dataset derived from web-based room tour videos that capture real-world indoor spaces and human walking demonstrations. Unlike existing VLN datasets, RoomTour3D leverages the scale and diversity of online videos to generate open-ended human walking trajectories and open-world navigable instructions. To compensate for the lack of navigation data in online videos, we perform 3D reconstruction and obtain 3D trajectories of walking paths augmented with additional information on the room types, object locations and 3D shape of surrounding scenes. Our dataset includes ~100K open-ended description-enriched trajectories with ~200K instructions, and 17K action-enriched trajectories from 1847 room tour environments. We demonstrate experimentally that RoomTour3D enables significant improvements across multiple VLN tasks including CVDN, SOON, R2R, and REVERIE. Moreover, Room-Tour3D facilitates the development of trainable zero-shot VLN agents, showcasing the potential and challenges of advancing towards open-world navigation.

GPT-4o, an omni-modal model that enables vocal conversations with diverse emotions and tones, marks a milestone for omni-modal foundation models. However, empowering Large Language Models to perceive and generate images, texts, and speeches end-to-end with publicly available data remains challenging for the open-source community. Existing vision-language models rely on external tools for speech processing, while speech-language models still suffer from limited or totally without vision-understanding capabilities. To address this gap, we propose the EMOVA (EMotionally Omni-present Voice Assistant), to enable Large Language Models with end-to-end speech abilities while maintaining the leading vision-language performance. With a semantic-acoustic disentangled speech tokenizer, we surprisingly notice that omni-modal alignment can further enhance vision-language and speech abilities compared with the bi-modal aligned counterparts. Moreover, a lightweight style module is introduced for the flexible speech style controls including emotions and pitches. For the first time, EMOVA achieves state-of-the-art performance on both the vision-language and speech benchmarks, and meanwhile, supporting omni-modal spoken dialogue with vivid emotions.

High-resolution image inputs allow Large Vision-Language Models (LVLMs) to capture finer visual details, improving comprehension. However, the increased training and computational costs associated with such inputs pose significant challenges. A common approach to mitigate these costs involves slicing the input into uniform patches using sliding windows, each aligned with the vision encoder’s input size. While efficient, this method fragments the input, disrupting the continuity of context, which negatively impacts cross-patch perception tasks. To address these limitations, we propose HiRes-LLaVA, a novel framework designed to efficiently process high-resolution inputs of any size without altering the original contextual and geometric information. HiRes-LLaVA introduces two key components: (i) a SliceRestore Adapter (SRA) that reconstructs sliced patches into their original form, enabling efficient extraction of both global and local features through down-up-sampling and convolutional layers, and (ii) a Self-Mining Sampler (SMS) that compresses visual tokens based on internal relationships, preserving original context and positional information while reducing training overhead. To assess the ability of handling context fragmentation, we construct a new benchmark, EntityGrid-QA, consisting of edge-related tasks. Extensive experiments demonstrate the superiority of HiRes-LLaVA on both existing public benchmarks and EntityGrid-QA. For example, with SRA, our method achieves a performance improvement of ∼ 12% over state-of-the-art LVLMs in addressing fragmentation issues. Additionally, our SMS outperforms other visual token downsamplers, while offering high data efficiency.

Currently, instruction-based image editing methods have made significant progress by leveraging the powerful cross-modal understanding capabilities of vision language models (VLMs). However, they still face challenges in three key areas: 1) complex scenarios; 2) semantic consistency; and 3) fine-grained editing. To address these issues, we propose FireEdit, an innovative Fine-grained Instruction-based image editing framework that exploits a REgion-aware VLM. FireEdit is designed to accurately comprehend user instructions and ensure effective control over the editing process. Specifically, we enhance the fine-grained visual perception capabilities of the VLM by introducing additional region tokens. Relying solely on the output of the LLM to guide the diffusion model may lead to suboptimal editing results. Therefore, we propose a Time-Aware Target Injection module and a Hybrid Visual Cross Attention module. The former dynamically adjusts the guidance strength at various denoising stages by integrating timestep embeddings with the text embeddings. The latter enhances visual details for image editing, thereby preserving semantic consistency between the edited result and the source image. By combining the VLM enhanced with fine-grained region tokens and the time-dependent diffusion model, FireEdit demonstrates significant advantages in comprehending editing instructions and maintaining high semantic consistency. Extensive experiments indicate that our approach surpasses the state-of-the-art instruction-based image editing methods.

Xiwen Liang, Min Lin, Weiqi Ruan, Rongtao Xu, Yuecheng Liu, Jiaqi Chen, Bingqian Lin, Yuzheng Zhuang, Xiaodan Liang. Proceedings of the 2025 Conference on Empirical Methods in Natural Language Processing. 2025.

The diffusion model is widely leveraged for either controllable video generation or video interpolation. As each field has its task-specific problems, it is difficult to merely develop a single model for completing both tasks simultaneously. Moreover, most existing works only support image conditions and necessitate redesigning the model structure to accommodate other types of conditions. Even so, they still face frame flickering issues when using the image as the condition due to the strong alignment of image pixels. To tackle these problems, in this work, we are the first to propose a unified diffusion framework, EasyControl, for both tasks of controllable video generation and interpolation with different types of conditions. The proposed EasyControl introduces a condition adapter to extract the condition features, which is then injected into an interchangeable fundamental text-to-video model to guide the video generation. To alleviate frame flicker problems, we propose a module named VideoInit to integrate the low-frequency band of input condition images, ensuring smoother generation. Experimental results on four benchmarks suggest that our method outperforms the previous methods on each task.

Image-to-video generation, which aims to generate a video starting from a given reference image, has drawn great attention. Existing methods frequently integrate semantic information from images or simply concatenate images, which often leads to low fidelity and flickering in the generated videos. To tackle these problems, we propose a high-fidelity image-to-video generation method by devising a frame retention branch based on a pre-trained video diffusion model, named DreamVideo. Our DreamVideo perceives the reference image via convolution layers and concatenates the features with the noisy latents as model input. By this means, the details of the reference image can be preserved to the greatest extent. In addition, by incorporating the designed double-condition classifier-free guidance, DreamVideo can generate high-quality videos of different actions by providing varying prompt texts. We conduct comprehensive experiments on the public datasets, and both quantitative and qualitative results indicate that our method outperforms the state-of-the-art method.

Operating robots in open-ended scenarios with diverse tasks is a crucial research and application direction in robotics. While recent progress in natural language processing and large multimodal models has enhanced robots' ability to understand complex instructions, robot manipulation still faces the procedural skill dilemma and the declarative skill dilemma in open environments. Existing methods often compromise cognitive and executive capabilities. To address these challenges, in this paper, we propose RoBridge, a hierarchical intelligent architecture for general robotic manipulation. It consists of a high-level cognitive planner (HCP) based on a large-scale pre-trained vision-language model (VLM), an invariant operable representation (IOR) serving as a symbolic bridge, and a generalist embodied agent (GEA). RoBridge maintains the declarative skill of VLM and unleashes the procedural skill of reinforcement learning, effectively bridging the gap between cognition and execution. RoBridge demonstrates significant performance improvements over existing baselines, achieving a 75% success rate on new tasks and an 83% average success rate in sim-to-real generalization using only five real-world data samples per task. This work represents a significant step towards integrating cognitive reasoning with physical execution in robotic systems, offering a new paradigm for general robotic manipulation.

Recent advancements in human motion synthesis have focused on specific types of motions, such as human-scene interaction, locomotion or human-human interaction, however, there is a lack of a unified system capable of generating a diverse combination of motion types. In response, we introduce *Sitcom-Crafter*, a comprehensive and extendable system for human motion generation in 3D space, which can be guided by extensive plot contexts to enhance workflow efficiency for anime and game designers. The system is comprised of eight modules, three of which are dedicated to motion generation, while the remaining five are augmentation modules that ensure consistent fusion of motion sequences and system functionality. Central to the generation modules is our novel 3D scene-aware human-human interaction module, which addresses collision issues by synthesizing implicit 3D Signed Distance Function (SDF) points around motion spaces, thereby minimizing human-scene collisions without additional data collection costs. Complementing this, our locomotion and human-scene interaction modules leverage existing methods to enrich the system's motion generation capabilities. Augmentation modules encompass plot comprehension for command generation, motion synchronization for seamless integration of different motion types, hand pose retrieval to enhance motion realism, motion collision revision to prevent human collisions, and 3D retargeting to ensure visual fidelity. Experimental evaluations validate the system's ability to generate high-quality, diverse, and physically realistic motions, underscoring its potential for advancing creative workflows. Code and demonstration videos can be found in the supplementary files.

Virtual try-on methods based on diffusion models achieve realistic effects but often require additional encoding modules, a large number of training parameters, and complex preprocessing, which increases the burden on training and inference. In this work, we re-evaluate the necessity of additional modules and analyze how to improve training efficiency and reduce redundant steps in the inference process. Based on these insights, we propose CatVTON, a simple and efficient virtual try-on diffusion model that transfers in-shop or worn garments of arbitrary categories to target individuals by concatenating them along spatial dimensions as inputs of the diffusion model. The efficiency of CatVTON is reflected in three aspects: (1) Lightweight network. CatVTON consists only of a VAE and a simplified denoising UNet, removing redundant image and text encoders as well as cross-attentions, and includes just 899.06M parameters. (2) Parameter-efficient training. Through experimental analysis, we identify self-attention modules as crucial for adapting pre-trained diffusion models to the virtual try-on task, enabling high-quality results with only 49.57M training parameters. (3) Simplified inference. CatVTON eliminates unnecessary preprocessing, such as pose estimation, human parsing, and captioning, requiring only a person image and garment reference to guide the virtual try-on process, reducing over 49% memory usage compared to other diffusion-based methods. Extensive experiments demonstrate that CatVTON achieves superior qualitative and quantitative results compared to baseline methods and demonstrates strong generalization performance in in-the-wild scenarios, despite being trained solely on public datasets with 73K samples.

Recent advancements in multi-modal 3D pre-training methods have shown promising efficacy in learning joint representations of text, images, and point clouds. However, adopting point clouds as 3D representation fails to fully capture the intricacies of the 3D world and exhibits a noticeable gap between the discrete points and the dense 2D pixels of images. To tackle this issue, we propose UniGS, integrating 3D Gaussian Splatting (3DGS) into multi-modal pre-training to enhance the 3D representation. We first rely on the 3DGS representation to model the 3D world as a collection of 3D Gaussians with color and opacity, incorporating all the information of the 3D scene while establishing a strong connection with 2D images. Then, to achieve Language-Image-3D pertaining, UniGS starts with a pretrained vision-language model to establish a shared visual and textual space through extensive real-world image-text pairs. Subsequently, UniGS employs a 3D encoder to align the optimized 3DGS with the Language-Image representations to learn unified multi-modal representations. To facilitate the extraction of global explicit 3D features by the 3D encoder and achieve better cross-modal alignment, we additionally introduce a novel Gaussian-Aware Guidance module that guides the learning of fine-grained representations of the 3D domain. Through extensive experiments across the Objaverse, ABO, MVImgNet and SUN RGBD datasets with zero-shot classification, text-driven retrieval and open-world understanding tasks, we demonstrate the effectiveness of UniGS in learning a more general and stronger aligned multi-modal representation. Specifically, UniGS achieves leading results across different 3D tasks with remarkable improvements over previous SOTA, Uni3D, including on zero-shot classification (+9.36%), text-driven retrieval (+4.3%) and open-world understanding (+7.92%).

Recent interactive point-based image manipulation methods have gained considerable attention for being user-friendly. However, these methods still face two types of ambiguity issues that can lead to unsatisfactory outcomes, namely, intention ambiguity which misinterprets the purposes of users, and content ambiguity where target image areas are distorted by distracting elements. To address these issues and achieve general-purpose manipulations, we propose a novel task-aware, training-free framework called GDrag. Specifically, GDrag defines a taxonomy of atomic manipulations, which can be parameterized and combined unitedly to represent complex manipulations, thereby reducing intention ambiguity. Furthermore, GDrag introduces two strategies to mitigate content ambiguity, including an anti-ambiguity dense trajectory calculation method (ADT) and a self-adaptive motion supervision method (SMS). Given an atomic manipulation, ADT converts the sparse user-defined handle points into a dense point set by selecting their semantic and geometric neighbors, and calculates the trajectory of the point set. Unlike previous motion supervision methods relying on a single global scale for low-rank adaption, SMS jointly optimizes point-wise adaption scales and latent feature biases. These two methods allow us to model fine-grained target contexts and generate precise trajectories. As a result, GDrag consistently produces precise and appealing results in different editing tasks. Extensive experiments on the challenging DragBench dataset demonstrate that GDrag outperforms state-of-the-art methods significantly. The code of GDrag will be released upon acceptance.

The global self-attention mechanism in diffusion transformers involves redundant computation due to the sparse and redundant nature of visual information, and the attention map of tokens within a spatial window shows significant similarity. To address this redundancy, we propose the Proxy-Tokenized Diffusion Transformer (PT-DiT), which employs sparse representative token attention (where the number of representative tokens is much smaller than the total number of tokens) to efficiently model global visual information. Specifically, within each transformer block, we compute an averaging token from each spatial-temporal window to serve as a proxy token for that region. The global semantics are captured through the self-attention of these proxy tokens and then injected into all latent tokens via cross-attention. Simultaneously, we introduce window and shift window attention to address the limitations in detail modeling caused by the sparse attention mechanism. Building on the well-designed PT-DiT, we further develop the PT-T2I/V family, which includes a variety of models for T2I, T2V, and T2MV tasks. Experimental results show that PT-DiT achieves competitive performance while reducing computational complexity in image and video generation tasks (e.g., a reduction 59\% compared to DiT and a reduction 34\% compared to PixArt-$\alpha$). The visual exhibition of and code are available at https://360cvgroup.github.io/Qihoo-T2X/.

Large language models (LLMs) have exhibited their problem-solving abilities in mathematical reasoning. Solving realistic optimization (OPT) problems in application scenarios requires advanced and applied mathematics ability. However, current OPT benchmarks that merely solve linear programming are far from complex realistic situations. In this work, we propose **OptiBench**, a benchmark for End-to-end optimization problem-solving with human-readable inputs and outputs. **OptiBench** contains rich optimization problems, including linear and nonlinear programming with or without tabular data, which can comprehensively evaluate LLMs' solving ability. In our benchmark, LLMs are required to call a code solver to provide precise numerical answers. Furthermore, to alleviate the data scarcity for optimization problems, and to bridge the gap between open-source LLMs on a small scale (e.g., Llama-3-8b) and closed-source LLMs (e.g., GPT-4), we further propose a data synthesis method namely ***ReSocratic***. Unlike general data synthesis methods that proceed from questions to answers, \ReSocratic first incrementally synthesizes formatted optimization demonstration with mathematical formulations step by step and then back-translates the generated demonstrations into questions. Based on this, we synthesize the ***ReSocratic-29k*** dataset. We further conduct supervised fine-tuning with ***ReSocratic-29k*** on multiple open-source models. Experimental results show that ***ReSocratic-29k*** significantly improves the performance of open-source models.

3D multiple object tracking (MOT) plays a crucial role in autonomous driving perception. Recent end-to-end query-based trackers simultaneously detect and track objects, which have shown promising potential for the 3D MOT task. However, existing methods are still in the early stages of development and lack systematic improvements, failing to track objects in certain complex scenarios, like occlusions and the small size of target object’s situations. In this paper, we first summarize the current end-to-end 3D MOT framework by decomposing it into three constituent parts: query initialization, query propagation, and query matching. Then we propose corresponding improvements, which lead to a strong yet simple tracker: S2-Track. Specifically, for query initialization, we present 2D-Prompted Query Initialization, which leverages predicted 2D object and depth information to prompt an initial estimate of the object’s 3D location. For query propagation, we introduce an Uncertainty-aware Probabilistic Decoder to capture the uncertainty of complex environment in object prediction with probabilistic attention. For query matching, we propose a Hierarchical Query Denoising strategy to enhance training robustness and convergence. As a result, our S2-Track achieves state-of-the-art performance on nuScenes benchmark, i.e., 66.3% AMOTA on test split, surpassing the previous best end-to-end solution by a significant margin of 8.9% AMOTA. We achieve 1st place on the nuScenes tracking task leaderboard.

Camera-based occupancy prediction is a main-stream approach for 3D perception in autonomous driving, aiming to infer complete 3D scene geometry and semantics from 2D images. Almost existing methods focus on improving performance through structural modifications, such as lightweight backbones and complex cascaded frameworks, with good yet limited performance. Few studies explore from the perspective of representation fusion, leaving the rich diversity of features in 2D images underutilized. Motivated by this, we propose CIGOcc, a two-stage occupancy prediction framework based on multi-level representation fusion. CIGOcc extracts segmentation, graphics, and depth features from an input image and introduces a deformable multi-level fusion mechanism to fuse these three multi-level features. Additionally, CIGOcc incorporates knowledge distilled from SAM to further enhance prediction accuracy. Without increasing training costs, CIGOcc achieves state-of-the-art performance on the SemanticKITTI benchmark. The code is provided in the supplementary material and will be released project page.

With the growing need for diverse and scalable data in indoor scene tasks, such as question answering and dense captioning, we propose 3D-MoRe, a novel paradigm designed to generate large-scale 3D-language datasets by lever-aging the strengths of foundational models. The framework integrates key components, including multi-modal embedding, cross-modal interaction, and a language model decoder, to process natural language instructions and 3D scene data. This approach facilitates enhanced reasoning and response generation in complex 3D environments. Using the ScanNet 3D scene dataset, along with text annotations from ScanQA and ScanRefer, 3D-MoRe generates 62,000 question-answer (QA) pairs and 73,000 object descriptions across 1,513 scenes. We also employ various data augmentation techniques and implement semantic filtering to ensure high-quality data. Experiments on ScanQA demonstrate that 3D-MoRe significantly outperforms state-of-the-art baselines, with the CIDEr score improving by 2.15%. Similarly, on ScanRefer, our approach achieves a notable increase in [email protected] by 1.84%, highlighting its effectiveness in both tasks. Our code and generated datasets will be publicly released to benefit the community, and both can be accessed on the https://3D-MoRe.github.io.

In this paper, we present LaVieID, a novel local a utoregressive vi deo diffusion framework designed to tackle the challenging id entity-preserving text-to-video task. The key idea of LaVieID is to mitigate the loss of identity information inherent in the stochastic global generation process of diffusion transformers (DiTs) from both spatial and temporal perspectives. Specifically, unlike the global and unstructured modeling of facial latent states in existing DiTs, LaVieID introduces a local router to explicitly represent latent states by weighted combinations of fine-grained local facial structures. This alleviates undesirable feature interference and encourages DiTs to capture distinctive facial characteristics. Furthermore, a temporal autoregressive module is integrated into LaVieID to refine denoised latent tokens before video decoding. This module divides latent tokens temporally into chunks, exploiting their long-range temporal dependencies to predict biases for rectifying tokens, thereby significantly enhancing inter-frame identity consistency. Consequently, LaVieID can generate high-fidelity personalized videos and achieve state-of-the-art performance. Our code and models are available at https://github.com/ssugarwh/LaVieID.

Autonomous driving has seen remarkable advancements, largely driven by extensive real-world data collection. However, acquiring diverse and corner-case data remains costly and inefficient. Generative models have emerged as a promising solution by synthesizing realistic sensor data. However, existing approaches primarily focus on single-modality generation, leading to inefficiencies and misalignment in multimodal sensor data. To address these challenges, we propose OminiGen, which generates aligned multimodal sensor data in a unified framework. Our approach leverages a shared Bird's Eye View (BEV) space to unify multimodal features and designs a novel generalizable multimodal reconstruction method, UAE, to jointly decode LiDAR and multi-view camera data. UAE achieves multimodal sensor decoding through volume rendering, enabling accurate and flexible reconstruction. Furthermore, we incorporate a Diffusion Transformer (DiT) with a ControlNet branch to enable controllable multimodal sensor generation. Our comprehensive experiments demonstrate that OminiGen achieves desired performances in unified multimodal sensor data generation with multimodal consistency and flexible sensor adjustments.

Haonan Wang, Minbin Huang, Runhui Huang, Lanqing Hong, Hang Xu, Tianyang Hu, Xiaodan Liang, Zhenguo Li, Hong Cheng, Kenji Kawaguchi. Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers). 2025.

Evaluating the step-by-step reliability of large language model (LLM) reasoning, such as Chain-of-Thought, remains challenging due to the difficulty and cost of obtaining high-quality step-level supervision. In this paper, we introduce Self-Play Critic (SPC), a novel approach where a critic model evolves its ability to assess reasoning steps through adversarial self-play games, eliminating the need for manual step-level annotation. SPC involves fine-tuning two copies of a base model to play two roles, namely a "sneaky generator" that deliberately produces erroneous steps designed to be difficult to detect, and a "critic" that analyzes the correctness of reasoning steps. These two models engage in an adversarial game in which the generator aims to fool the critic, while the critic model seeks to identify the generator's errors. Using reinforcement learning based on the game outcomes, the models iteratively improve; the winner of each confrontation receives a positive reward and the loser receives a negative reward, driving continuous self-evolution. Experiments on three reasoning process benchmarks (ProcessBench, PRM800K, DeltaBench) demonstrate that our SPC progressively enhances its error detection capabilities (e.g., accuracy increases from 70.8% to 77.7% on ProcessBench) and surpasses strong baselines, including distilled R1 model. Furthermore, SPC can guide the test-time search of diverse LLMs and significantly improve their mathematical reasoning performance on MATH500 and AIME2024, surpassing those guided by state-of-the-art process reward models.

While vision-language models (VLMs) have demonstrated promising capabilities in reasoning and planning for embodied agents, their ability to comprehend physical phenomena, particularly within structured 3D environments, remains severely limited. To close this gap, we introduce PhyBlock, a progressive benchmark designed to assess VLMs on physical understanding and planning through robotic 3D block assembly tasks. PhyBlock integrates a novel four-level cognitive hierarchy assembly task alongside targeted Visual Question Answering (VQA) samples, collectively aimed at evaluating progressive spatial reasoning and fundamental physical comprehension, including object properties, spatial relationships, and holistic scene understanding. PhyBlock includes 2600 block tasks (400 assembly tasks, 2200 VQA tasks) and evaluates models across three key dimensions: partial completion, failure diagnosis, and planning robustness. We benchmark 23 state-of-the-art VLMs, highlighting their strengths and limitations in physically grounded, multi-step planning. Our empirical findings indicate that the performance of VLMs exhibits pronounced limitations in high-level planning and reasoning capabilities, leading to a notable decline in performance for the growing complexity of the tasks.Error analysis reveals persistent difficulties in spatial orientation and dependency reasoning.We position PhyBlock as a unified testbed to advance embodied reasoning, bridging vision-language understanding and real-world physical problem-solving.

Recent advances in text-to-video (T2V) generation, exemplified by models such as Sora and Kling, have demonstrated strong potential for constructing world simulators. However, existing T2V models still struggle to understand abstract physical principles and to generate videos that faithfully obey physical laws. This limitation stems primarily from the lack of explicit physical guidance, caused by a significant gap between high-level physical concepts and the generative capabilities of current models. To address this challenge, we propose the **W**orld **S**imulator **A**ssistant (**WISA**), a novel framework designed to systematically decompose and integrate physical principles into T2V models. Specifically, WISA decomposes physical knowledge into three hierarchical levels: textual physical descriptions, qualitative physical categories, and quantitative physical properties. It then incorporates several carefully designed modules—such as Mixture-of-Physical-Experts Attention (MoPA) and a Physical Classifier—to effectively encode these attributes and enhance the model’s adherence to physical laws during generation. In addition, most existing video datasets feature only weak or implicit representations of physical phenomena, limiting their utility for learning explicit physical principles. To bridge this gap, we present **WISA-80K**, a new dataset comprising 80,000 human-curated videos that depict 17 fundamental physical laws across three core domains of physics: dynamics, thermodynamics, and optics. Experimental results show that WISA substantially improves the alignment of T2V models (such as CogVideoX and Wan2.1) with real-world physical laws, achieving notable gains on the VideoPhy benchmark. Our data, code, and models are available in the [Project Page](https://wisav1.github.io/WISA/).

Spatial Planning is a crucial part in the field of spatial intelligence, which requires the understanding and planning about object arrangements in space perspective. AI agents with the spatial planning ability can better adapt to various real-world applications, including robotic manipulation, automatic assembly, urban planning etc. Recent works have attempted to construct benchmarks for evaluating the spatial intelligence of Multimodal Large Language Models (MLLMs). Nevertheless, these benchmarks primarily focus on spatial reasoning based on typical Visual Question-Answering (VQA) forms, which suffers from the gap between abstract spatial understanding and concrete task execution. In this work, we take a step further to build a comprehensive benchmark called MineAnyBuild, aiming to evaluate the spatial planning ability of open-world AI agents in the Minecraft game. Specifically, MineAnyBuild requires an agent to generate executable architecture building plans based on the given multi-modal human instructions. It involves 4,000 curated spatial planning tasks and also provides a paradigm for infinitely expandable data collection by utilizing rich player-generated content. MineAnyBuild evaluates spatial planning through four core supporting dimensions: spatial understanding, spatial reasoning, creativity, and spatial commonsense. Based on MineAnyBuild, we perform a comprehensive evaluation for existing MLLM-based agents, revealing the severe limitations but enormous potential in their spatial planning abilities. We believe our MineAnyBuild will open new avenues for the evaluation of spatial intelligence and help promote further development for open-world AI agents capable of spatial planning.

We present SeePhys, a large-scale multimodal benchmark for LLM reasoning grounded in physics questions ranging from middle school to PhD qualifying exams. The benchmark covers 7 fundamental domains spanning the physics discipline, incorporating 21 categories of highly heterogeneous diagrams. In contrast to prior works where visual elements mainly serve auxiliary purposes, our benchmark features a substantial proportion of vision-essential problems (75%) that mandate visual information extraction for correct solutions. Through extensive evaluation, we observe that even the most advanced visual reasoning models (e.g., Gemini-2.5-pro and o4-mini) achieve sub-60% accuracy on our benchmark. These results reveal fundamental challenges in current large language models' visual understanding capabilities, particularly in: (i) establishing rigorous coupling between diagram interpretation and physics reasoning, and (ii) overcoming their persistent reliance on textual cues as cognitive shortcuts.Project Page: github.com/SeePhys/seephys-projectHugging Face: huggingface.co/datasets/SeePhys/SeePhys

Despite promising progress in face swapping task, realistic swapped images remain elusive, often marred by artifacts, particularly in scenarios involving high pose variation, color differences, and occlusion. To address these issues, we propose a novel approach that better harnesses diffusion models for face-swapping by making following core contributions. (a) We propose to reframe the face-swapping task as a self-supervised, train-time inpainting problem, enhancing the identity transfer while blending with the target image. (b) We introduce a multi-step De-noising Diffusion Implicit Model (DDIM) sampling during training, reinforcing identity and perceptual similarities. (c) Third, we introduce CLIP feature disentanglement to extract pose, expression, and lighting information from the target image, improving fidelity. (d) Further, we introduce a mask shuffling technique during inpainting training, which allows us to create a so-called universal model for swapping, with an additional feature of head swapping. Ours can swap hair and even accessories, beyond traditional face swapping. Unlike prior works reliant on multiple off-the-shelf models, ours is a relatively unified approach and so it is resilient to errors in other off-the-shelf models. Extensive experiments on FFHQ and CelebA datasets validate the efficacy and robustness of our approach, show-casing high-fidelity, realistic face-swapping with minimal inference time. Our code is available at REFace.

Actionable part perception for robotic object manipulation needs to perceive parts over open-world object categories within 3D space, which is challenging as the appearance of the same part on different objects varies greatly. It is frequently observed that despite the huge intra-class difference in appearance, the parts share common interactive functions over different objects, i.e., common affordance. According to this observation, we propose DisCo, a novel technique that Discovers Common affordance information from powerful large models for guiding the actionable part perception across open-world objects. Specifically, we first use a large language model to identify the object names that each part potentially belongs to and a text-to-image generative model to generate image examples for the queried objects, constructing image-text paired data that indicate visual and semantic information of common affordance. Then, our model encodes the common affordance information by learning to pair the object-part images with their text descriptions. Subsequently, the 2D-pixel features are distilled into 3D space, thus the 3D point features are enriched with not only the semantic information of open-set objects but also the common affordance information which is highly generalizable. Finally, a segmentation head and a pose regression network are developed to predict more accurate results of part segmentation and pose estimation, improving the success rate of robotic object manipulation. Extensive experiments show that our method outperforms existing methods on the part instance and semantic segmentation by significant margins of 4.8% mAp, 5.4% AP50, and 3.9% mIoU on the unseen object categories.

Recent advances in hand-crafted neural architectures for visual recognition underscore the pressing need to explore architecture designs comprising diverse building blocks. Concurrently, neural architecture search (NAS) methods have gained traction as a means to alleviate human efforts. Nevertheless, the question of whether NAS methods can efficiently and effectively manage diversified search spaces featuring disparate candidates, such as Convolutional Neural Networks (CNNs) and transformers, remains an open question. In this work, we introduce a novel unsupervised NAS approach called BossNAS (Block-wisely Self-supervised Neural Architecture Search), which aims to address the problem of inaccurate predictive architecture ranking caused by a large weight-sharing space while mitigating potential ranking issue caused by biased supervision. To achieve this, we factorize the search space into blocks and introduce a novel self-supervised training scheme called Ensemble Bootstrapping, to train each block separately in an unsupervised manner. In the search phase, we propose an unsupervised Population-Centric Search, optimizing the candidate architecture towards the population center. Additionally, we enhance our NAS method by integrating masked image modeling and present BossNAS++ to overcome the lack of dense supervision in our block-wise self-supervised NAS. In BossNAS++, we introduce the training technique named Masked Ensemble Bootstrapping for block-wise supernet, accompanied by a Masked Population-Centric Search scheme to promote fairer architecture selection. Our family of models, discovered through BossNAS and BossNAS++, delivers impressive results across various search spaces and datasets. Our transformer model discovered by BossNAS++ attains a remarkable accuracy of 83.2% on ImageNet with only 10.5B MAdds, surpassing DeiT-B by 1.4% while maintaining a lower computation cost. Moreover, our approach excels in architecture rating accuracy, achieving Spearman correlations of 0.78 and 0.76 on the canonical MBConv search space with ImageNet and the NATS-Bench size search space with CIFAR-100, respectively, outperforming state-of-the-art NAS methods.

Vision-and-Language Navigation (VLN), as a crucial research problem of Embodied AI, requires an embodied agent to navigate through complex 3D environments following natural language instructions. Recent research has highlighted the promising capacity of large language models (LLMs) in VLN by improving navigational reasoning accuracy and interpretability. However, their predominant use in an offline manner usually suffers from substantial domain gap between the VLN task and the LLM training corpus. This paper proposes a novel strategy called Navigational Chain-of-Thought (NavCoT), where we fulfill parameter-efficient in-domain training to enable self-guided navigational decision, leading to a significant mitigation of the domain gap in a cost-effective manner. Specifically, at each timestep, the LLM is prompted to forecast the navigational chain-of-thought by: 1) acting as a world model to imagine the next observation according to the instruction, 2) selecting the candidate observation that best aligns with the imagination, and 3) determining the action based on the reasoning from the prior steps. In this way, the action prediction can be effectively simplified benefiting from the disentangled reasoning. Through constructing formalized labels for training, the LLM can learn to generate desired and reasonable chain-of-thought outputs for improving the action decision. Experimental results across various training settings and popular VLN benchmarks (e.g., Room-to-Room (R2R), Room-across-Room (RxR), Room-for-Room (R4R)) show the significant superiority of NavCoT over the direct action prediction variants. Through simple parameter-efficient finetuning, our NavCoT outperforms a recent GPT4-based approach with $\sim$∼7% relative improvement on the R2R dataset. We believe that NavCoT will help unlock more task-adaptive and scalable LLM-based embodied agents, which are helpful for developing real-world robotics applications.

We introduce a monocular-to-3D virtual try-on network based on a conditional 3D-aware Generative Adversarial Network (3D-GAN) for synthesizing multi-view try-on results from single monocular images. In contrast to previous 3D virtual try-on methods that rely on costly scanned meshes or pseudo-depth maps for supervision, our approach utilizes a conditional 3D-GAN trained solely on 2D images, greatly simplifying dataset construction and enhancing model scalability. Specifically, we propose a Generative monocular-to-3D Virtual Try-ON network (G3D-VTON) that integrates a 3D-aware conditional Parsing Module (3DPM), a U-Net Refinement Module (URM), and a Flow-based 2D Virtual Try-On Module (FTM). In our framework, the 3DPM is designed to generate a 3D representation of the virtual try-on result, thereby enabling multi-view rendering. To accomplish this, it is implemented using conditional generative semantic articulated fields, which leverage the 3D SMPL prior via inverse skinning to learn the Signed Distance Function (SDF) of the try-on results in a canonical pose space. This learned SDF enables the rendering of both a coarse human parsing map and a preliminary try-on output with explicit camera control. Furthermore, within 3DPM, we introduce deferred pose guidance to decouple style and pose conditions during training, thereby facilitating view controllable generation during inference. However, the rendered human parsing and try-on results exhibit imprecise shapes and blurry textures. To address these issues, the URM subsequently refines these rendered outputs using a refinement U-Net, and the FTM integrates the refined results with the 2D warped garment to generate the final try-on output with more accurate and realistic appearance details. Extensive experiments demonstrate that the proposed G3D-VTON effectively manipulates and generates faithful 3D human appearances wearing the desired garment, outperforming both 3D-GAN and depth-based 3D approaches while delivering superior visual results in 2D.

Detecting foreground objects in video separation tasks is a challenging endeavor in complex environments. This paper presents a deep neural network architecture considering the features of spectral, spatial, and temporal at the same time. Under this framework, we designed a gated mechanism for attention and incorporated it into the Transformer architecture (GMAT). This model employs a gating mechanism to dynamically control and allocate attention across different features (wavelet features and raw features), learning how to balance their importance, emphasize or ignore certain features based on the current context. Additionally, in order to enhance features in video frame data and better focus on important features while ignoring unimportant ones in GMAT, we introduced an optical flow estimation method based on wavelet transform. Due to the advantage of wavelet transform in capturing motion features at different scales, its introduction enables a more comprehensive and detailed focus on finer features. We evaluated the GMAT on various videos from the CDNet2014 dataset, results of qualitative and quantitative evaluations demonstrating its significant improved for foreground detection over several recent video separation models.

Condition-based video generation aims to create video content based on given information that describes specific subjects. However, most existing works can only utilize a single condition to guide the denoising process, thereby limiting their applicability to specific scenarios. Although some works attempt to accommodate multiple conditions within one framework, they often require multiple encoders, leading to inefficiencies in integrating multi-condition features. In this work, we present a framework that, with the support of the proposed Unified Adapter (UniAdapter), enables simultaneous multi-condition control of video generation within a single model. To effectively merge these conditions, we propose a novel Probabilistic Multi-condition Concatenator (PMC) module, which employs a unified encoder to accommodate multiple conditions and concatenate condition features at the pixel level to achieve fine-grained control. Following the PMC module, we employ 2D down-sampling blocks to refine features for injection into the Video Diffusion Model (VDM). Moreover, our UniAdapter is designed to be model-agnostic and compatible with any U-Net-based VDM, offering a versatile solution for improving video generation quality. Experimental results on public benchmarks UCF-101 and MSR-VTT show that our method achieves superior results in both quantitative and qualitative evaluations.

The pre-trained vision-language model, exemplified by CLIP, advances zero-shot semantic segmentation by aligning visual features with class embeddings through a transformer decoder to generate semantic masks. Despite its effectiveness, prevailing methods within this paradigm encounter challenges, including overfitting on seen classes and small fragmentation in segmentation masks. To mitigate these issues, we propose a Language-Driven Visual Consensus (LDVC) approach, fostering improved alignment of linguistic and visual information. Specifically, we leverage class embeddings as anchors due to their discrete and abstract nature, steering visual features toward class embeddings. Moreover, to achieve a more compact visual space, we introduce route attention into the transformer decoder to find visual consensus, thereby enhancing semantic consistency within the same object. Equipped with a vision-language prompting strategy, our approach significantly boosts the generalization capacity of segmentation models for unseen classes. Experimental results underscore the effectiveness of our approach, showcasing mIoU gains of 4.5% on the PASCAL VOC 2012 and 3.6% on the COCO-Stuff 164K for unseen classes compared with the state-of-the-art methods.

Recent advances in text-to-image diffusion models have achieved remarkable success in generating high-quality, realistic images from textual descriptions. However, these approaches have faced challenges in precisely aligning the generated visual content with the textual concepts described in the prompts. In this article, we propose a two-stage coarse-to-fine semantic realignment method, named RealignDiff, aimed at improving the alignment between text and images in text-to-image diffusion models. In the coarse semantic realignment phase, a novel caption reward, leveraging the BLIP-2 model, is proposed to evaluate the semantic discrepancy between the generated image caption and the given text prompt. Subsequently, the fine semantic realignment stage uses a local dense caption generation module and a reweighting attention modulation module to refine the previously generated images from a local semantic view. Experimental results on the MS-COCO and ViLG-300 datasets demonstrate that the proposed two-stage coarse-to-fine semantic realignment method outperforms other baseline realignment techniques by a substantial margin in both visual quality and semantic similarity with the input prompt.

Considering how to make the model accurately understand and follow natural language instructions and perform actions consistent with world knowledge is a key challenge in robot manipulation. This mainly includes human fuzzy instruction reasoning and the following of physical knowledge. Therefore, the embodied intelligence agent must have the ability to model world knowledge from training data. However, most existing vision and language robot manipulation methods mainly operate in less realistic simulators and language settings and lack explicit modeling of world knowledge. To bridge this gap, we introduce a novel and simple robot manipulation framework, called Surfer. It is based on the world model, treats robot manipulation as a state transfer of the visual scene, and decouples it into two parts: action and scene. Then, the generalization ability of the model on new instructions and new scenes is enhanced by explicit modeling of the action and scene prediction in multimodal information. In addition, we built a robot manipulation simulation platform that supports physics execution based on the MuJoCo physics engine. It can automatically generate demonstration training data and test data, effectively reducing labor costs. To conduct a comprehensive and systematic evaluation of the visual-language understanding and physical execution of the manipulation model, we also created a robotic manipulation benchmark with different difficulty levels, called SeaWave. It contains four visual-language manipulation tasks of different difficulty levels and can provide a standardized testing platform for embedded AI agents in multimodal environments. Overall, we hope Surfer can freely surf in the robot's SeaWave benchmark. Extensive experiments show that Surfer consistently outperforms all baselines significantly in all manipulation tasks. On average, Surfer achieved a success rate of 54.74% on the defined four levels of manipulation tasks, exceeding the best baseline performance of 51.07%. The simulator, code, and benchmarks are released at https://pzhren.github.io/Surfer.

Neural radiance fields (NeRFs) are promising 3D representations for scenes, objects, and humans. However, most existing methods require multi-view inputs and per-scene training, which limits their real-life applications. Moreover, current methods focus on single-subject cases, leaving scenes of interacting hands that involve severe inter-hand occlusions and challenging view variations remain unsolved. To tackle these issues, this paper proposes a generalizable visibility-aware NeRF (VA-NeRF) framework for interacting hands. Specifically, given an image of interacting hands as input, our VA-NeRF first obtains a mesh-based representation of hands and extracts their corresponding geometric and textural features. Subsequently, a feature fusion module that exploits the visibility of query points and mesh vertices is introduced to adaptively merge features of both hands, enabling the recovery of features in unseen areas. Additionally, our VA-NeRF is optimized together with a novel discriminator within an adversarial learning paradigm. In contrast to conventional discriminators that predict a single real/fake label for the synthesized image, the proposed discriminator generates a pixel-wise visibility map, providing fine-grained supervision for unseen areas and encouraging the VA-NeRF to improve the visual quality of synthesized images. Experiments on the Interhand2.6M dataset demonstrate that our proposed VA-NeRF outperforms conventional NeRFs significantly. Project Page: https://github.com/XuanHuang0/VANeRF.

Current parametric models have made notable progress in 3D hand pose and shape estimation. However, due to the fixed hand topology and complex hand poses, current models are hard to generate meshes that are aligned with the image well. To tackle this issue, we introduce a dual noise estimation method in this paper. Given a single-view image as input, we first adopt a baseline parametric regressor to obtain the coarse hand meshes. We assume the mesh vertices and their image-plane projections are noisy, and can be associated in a unified probabilistic model. We then learn the distributions of noise to refine mesh vertices and their projections. The refined vertices are further utilized to refine camera parameters in a closed-form manner. Consequently, our method obtains well-aligned and high-quality 3D hand meshes. Extensive experiments on the large-scale Interhand2.6M dataset demonstrate that the proposed method not only improves the performance of its baseline by more than 10% but also achieves state-of-the-art performance. Project page: https://github.com/hanhuili/DNE4Hand.

Talking upper-body synthesis is a promising task due to its versatile potential for video creation and consists of animating the body and face from a source image with the motion from a given driving video. However, prior synthesis approaches fall short in addressing this task and have been either limited to animating heads of a target person only, or have animated the upper body but neglected the synthesis of precise facial details. To tackle this task, we propose a Photo-realistic Talking Upper-body Synthesis method via 3D-aware motion decomposition warping, named PTUS, to both precisely synthesize the upper body as well as recover the details of the face such as blinking and lip synchronization. In particular, the motion decomposition mechanism consists of a face-body motion decomposition, which decouples the 3D motion estimation of the face and body, and a local-global motion decomposition, which decomposes the 3D face motion into global and local motions resulting in the transfer of facial expression. The 3D-aware warping module transfers the large-scale and subtle 3D motions to the extracted 3D depth-aware features in a coarse-tofine manner. Moreover, we present a new dataset, Talking-UB, which includes upper-body images with high-resolution faces, addressing the limitations of prior datasets that either consist of only facial images or upper-body images with blurry faces. Experimental results demonstrate that our proposed method can synthesize high-quality videos that preserve facial details, and achieves superior results compared to state-of-the-art cross-person motion transfer approaches. Code and collected dataset are released in https://github.com/cooluoluo/PTUS.

Text-guided motion synthesis aims to generate 3D human motion that not only precisely reflects the textual description but reveals the motion details as much as possible. Pioneering methods explore the diffusion model for text-to-motion synthesis and obtain significant superiority. However, these methods conduct diffusion processes either on the raw data distribution or the low-dimensional latent space, which typically suffer from the problem of modality inconsistency or detail-scarce. To tackle this problem, we propose a novel Basic-to-Advanced Hierarchical Diffusion Model, named B2A-HDM, to collaboratively exploit low-dimensional and high-dimensional diffusion models for high quality detailed motion synthesis. Specifically, the basic diffusion model in low-dimensional latent space provides the intermediate denoising result that to be consistent with the textual description, while the advanced diffusion model in high-dimensional latent space focuses on the following detail-enhancing denoising process. Besides, we introduce a multi-denoiser framework for the advanced diffusion model to ease the learning of high-dimensional model and fully explore the generative potential of the diffusion model. Quantitative and qualitative experiment results on two text-to-motion benchmarks (HumanML3D and KIT-ML) demonstrate that B2A-HDM can outperform existing state-of-the-art methods in terms of fidelity, modality consistency, and diversity.

Despite the significant success of large visionlanguage models (LVLMs), some studies have revealed that LVLMs suffer from the hallucination problem, where the LVLMs' response contains descriptions of non-existent objects.Although various benchmarks have been proposed to investigate this problem, they mostly focus on single-turn evaluation and overlook the hallucination raised by textual inputs.To investigate the hallucination problem of LVLMs when given long-term misleading textual history, we propose a novel visual dialogue hallucination evaluation benchmark VisDiaHalBench.The benchmark consists of samples with five-turn questions about an edited image and its original version.Vis-DiaHalBench differs from previous hallucination benchmarks in the following three points: 1) The questions and answers are unambiguously grounded by annotated scene graphs.2) The images are uncommonly edited to inspect the visual model and common-object hallucination in LLMs.3) The carefully designed dialogue refers a same object in different turns to assess the image consistency and influence of history for LVLMs.The detailed analysis of several state-of-the-art LVLMs across image consistency, visual understanding, history influence, and other dimensions reveals their substantial performance gap with single-turn VQA tasks.The benchmark is released in: https://github.com/ qingxingcao/VisDiaHalBench

Embodied agents equipped with GPT as their brains have exhibited extraordinary decisionmaking and generalization abilities across various tasks.However, existing zero-shot agents for vision-and-language navigation (VLN) only prompt GPT-4 to select potential locations within localized environments, without constructing an effective "global-view" for the agent to understand the overall environment.In this work, we present a novel map-guided GPT-based agent, dubbed MapGPT, which introduces an online linguistic-formed map to encourage global exploration.Specifically, we build an online map and incorporate it into the prompts that include node information and topological relationships, to help GPT understand the spatial environment.Benefiting from this design, we further propose an adaptive planning mechanism to assist the agent in performing multi-step path planning based on a map, systematically exploring multiple candidate nodes or sub-goals step by step.Extensive experiments demonstrate that our MapGPT is applicable to both GPT-4 and GPT-4V, achieving state-of-the-art zero-shot performance on R2R and REVERIE simultaneously (∼10% and ∼12% improvements in SR), and showcasing the newly emergent global thinking and path planning abilities of the GPT.

Yinya Huang, Ruixin Hong, Hongming Zhang, Wei Shao, Zhicheng Yang, Dong Yu, Changshui Zhang, Xiaodan Liang, Linqi Song. Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2024.

Empathetic response generation is designed to comprehend the emotions of others and select the most appropriate strategies to assist them in resolving emotional challenges. Empathy can be categorized into cognitive empathy and affective empathy. The former pertains to the ability to understand and discern the emotional issues and situations of others, while the latter involves the capacity to provide comfort. To enhance one's empathetic abilities, it is essential to develop both these aspects. Therefore, we develop an innovative framework that combines retrieval augmentation and emotional support strategy integration. Our framework starts with the introduction of a comprehensive emotional palette for empathy. We then apply appraisal theory to decompose this palette and create a database of empathetic responses. This database serves as an external resource and enhances the LLM's empathy by integrating semantic retrieval mechanisms. Moreover, our framework places a strong emphasis on the proper articulation of response strategies. By incorporating emotional support strategies, we aim to enrich the model's capabilities in both cognitive and affective empathy, leading to a more nuanced and comprehensive empathetic response. Finally, we extract datasets ED and ET from the empathetic dialogue dataset EmpatheticDialogues and ExTES based on dialogue length. Experiments demonstrate that our framework can enhance the empathy ability of LLMs from both cognitive and affective empathy perspectives. Our code is released at https://github.com/CAS-SIAT-XinHai/APTNESS.

The rise of multimodal large language models (MLLMs) has spurred interest in language-based driving tasks. However, existing research typically focuses on limited tasks and often omits key multi-view and temporal information which is crucial for robust autonomous driving. To bridge these gaps, we introduce NuInstruct, a novel dataset with 91K multi-view video-QA pairs across 17 subtasks, where each task demands holistic information (e.g., temporal, multi-view, and spatial), significantly elevating the challenge level. To obtain NuInstruct, we propose a novel SQL-based method to generate instruction-response pairs automatically, which is inspired by the driving logical progression of humans. We further present BEV-InMLLM, an end-to-end method for efficiently deriving instruction-aware Bird's-Eye-View (BEV) features, language-aligned for large language models. BEV-InMLLM integrates multi-view, spatial awareness, and temporal semantics to enhance MLLMs' capabilities on NuInstruct tasks. Moreover, our proposed BEV injection module is a plug-and-play method for existing MLLMs. Our experiments on NuInstruct demonstrate that BEV-InMLLM significantly outperforms existing MLLMs, e.g. 9% improvement on various tasks. We release our NuInstruct at https://github.com/xmed-lab/NuInstruct.

Self-attention mechanism is the key of the Transformer but often criticized for its computation demands. Previous token pruning works motivate their methods from the view of computation redundancy but still need to load the full network and require same memory costs. This paper introduces a novel strategy that simplifies vision transformers and reduces computational load through the selective removal of non-essential attention layers, guided by entropy considerations. We identify that regarding the attention layer in bottom blocks, their subsequent MLP layers, i.e. two feed-forward layers, can elicit the same entropy quantity. Meanwhile, the accompanied MLPs are under-exploited since they exhibit smaller feature entropy compared to those MLPs in the top blocks. Therefore, we propose to integrate the uninformative attention layers into their subsequent counterparts by degenerating them into identical mapping, yielding only MLP in certain transformer blocks. Experimental results on ImageNet-1k show that the proposed method can remove 40% attention layer of DeiT-B, improving throughput and memory bound without performance compromise.

Neural implicit fields have been a de facto standard in novel view synthesis. Recently, there exist some methods exploring fusing multiple modalities within a single field, aiming to share implicit features from different modalities to enhance reconstruction performance. However, these modalities often exhibit misaligned behaviors: optimizing for one modality, such as LiDAR, can adversely affect another, like camera performance, and vice versa. In this work, we conduct comprehensive analyses on the multimodal implicit field of LiDAR-camera joint synthesis, revealing the underlying issue lies in the misalignment of different sensors. Furthermore, we introduce AlignMiF, a geometrically aligned multimodal implicit field with two proposed modules: Geometry-Aware Alignment (GAA) and Shared Geometry Initialization (SGI). These modules effectively align the coarse geometry across different modalities, significantly enhancing the fusion process between LiDAR and camera data. Through extensive experiments across various datasets and scenes, we demonstrate the effectiveness of our approach in facilitating better interaction between LiDAR and camera modalities within a unified neural field. Specifically, our proposed AlignMiF, achieves remarkable improvement over recent implicit fusion methods (+2.01 and +3.11 image PSNR on the KITTI-360 and Waymo datasets) and consistently surpasses single modality performance (13.8% and 14.2% reduction in LiDAR Cham-fer Distance on the respective datasets). Code release: https://github.com/tangtaogo/alignmif.

Existing open-vocabulary object detectors typically require a predefined set of categories from users, signifi-cantly confining their application scenarios. In this pa-per, we introduce DetCLIPv3, a high-performing detector that excels not only at both open-vocabulary object detection, but also generating hierarchical labels for detected objects. DetCLIPv3 is characterized by three core designs: 1. Versatile model architecture: we derive a robust open-set detection framework which is further empowered with generation ability via the integration of a caption head. 2. High information density data: we develop an auto-annotation pipeline leveraging visual large language model to refine captions for large-scale image-text pairs, providing rich, multi-granular object labels to enhance the training. 3. Efficient training strategy: we employ a pre-training stage with low-resolution inputs that enables the object captioner to efficiently learn a broad spectrum of visual concepts from extensive image-text paired data. This is followed by a fine-tuning stage that leverages a small number of high-resolution samples to further enhance detection performance. With these effective designs, DetCLIPv3 demonstrates superior open-vocabulary detection performance, e.g., our Swin- T backbone model achieves a notable 47.0 zero-shot fixed AP on the LVIS minival benchmark, outperforming GLIPv2, GroundingDINO, and DetCLIPv2 by 18.0/19.6/6.6Ap, respectively. DetCLIPv3 also achieves a state-of-the-art 19.7 AP in dense captioning task on VG dataset, showcasing its strong generative capability.

Recent large language models (LLMs) have witnessed significant advancement in various tasks, including mathematical reasoning and theorem proving. As these two tasks require strict and formal multi-step inference, they are appealing domains for exploring the reasoning ability of LLMs but still face important challenges. Previous studies such as Chain-of-Thought (CoT) have revealed the effectiveness of intermediate steps guidance. However, such step-wise annotation requires heavy labor, leading to insufficient training steps for current benchmarks. To fill this gap, this work introduces MUSTARD, a data generation framework that masters uniform synthesis of theorem and proof data of high quality and diversity. MUSTARD synthesizes data in three stages: (1) It samples a few mathematical concept seeds as the problem category. (2) Then, it prompts a generative language model with the sampled concepts to obtain both the problems and their step-wise formal solutions. (3) Lastly, the framework utilizes a proof assistant (e.g., Lean Prover) to filter the valid proofs. With the proposed MUSTARD, we present a theorem-and-proof benchmark MUSTARDSAUCE with 5,866 valid data points. Each data point contains an informal statement, an informal proof, and a translated formal proof that passes the prover validation. We perform extensive analysis and demonstrate that MUSTARD generates validated high-quality step-by-step data. We further apply the MUSTARDSAUCE for fine-tuning smaller language models. The fine-tuned Llama 2-7B achieves a 15.41% average relative performance gain in automated theorem proving, and 8.18% in math word problems. Codes and data are available at https://github.com/Eleanor-H/MUSTARD.

This paper introduces Instruction-oriented Object Detection (IOD), a new task that enhances human-computer interaction by enabling object detectors to understand user instructions and locate relevant objects. Unlike traditional open-vocabulary object detection tasks that rely on users providing a list of required category names, IOD requires models to comprehend natural-language instructions, contextual reasoning, and output the name and location of the desired categories. This poses fresh challenges for modern object detection systems. To develop an IOD system, we create a dataset called IOD-Bench, which consists of instruction-guided detections, along with specialized evaluation metrics. We leverage large-scale language models (LLMs) to generate a diverse set of instructions (8k+) based on existing public object detection datasets, covering a wide range of real-world scenarios. As an initial approach to the IOD task, we propose a model called Ins-DetCLIP. It harnesses the extensive knowledge within LLMs to empower the detector with instruction-following capabilities. Specifically, our Ins-DetCLIP employs a visual encoder (i.e., DetCLIP, an open-vocabulary detector) to extract object-level features. These features are then aligned with the input instructions using a cross-modal fusion module integrated into a pre-trained LLM. Experimental results conducted on IOD-Bench demonstrate that our model consistently outperforms baseline methods that directly combine LLMs with detection models. This research aims to pave the way for a more adaptable and versatile interaction paradigm in modern object detection systems, making a significant contribution to the field.

Despite the success of large language models (LLMs), the task of theorem proving still remains one of the hardest reasoning tasks that is far from being fully solved. Prior methods using language models have demonstrated promising results, but they still struggle to prove even middle school level theorems. One common limitation of these methods is that they assume a fixed theorem library during the whole theorem proving process. However, as we all know, creating new useful theorems or even new theories is not only helpful but crucial and necessary for advancing mathematics and proving harder and deeper results. In this work, we present LEGO-Prover, which employs a growing skill library containing verified lemmas as skills to augment the capability of LLMs used in theorem proving. By constructing the proof modularly, LEGO-Prover enables LLMs to utilize existing skills retrieved from the library and to create new skills during the proving process. These skills are further evolved (by prompting an LLM) to enrich the library on another scale. Modular and reusable skills are constantly added to the library to enable tackling increasingly intricate mathematical problems. Moreover, the learned library further bridges the gap between human proofs and formal proofs by making it easier to impute missing steps. LEGO-Prover advances the state-of-the-art pass rate on miniF2F-valid (48.0\% to 57.0\%) and miniF2F-test (45.5\% to 50.0\%). During the proving process, LEGO-Prover also generates over 20,000 skills (theorems/lemmas) and adds them to the growing library. Our ablation study indicates that these newly added skills are indeed helpful for proving theorems, resulting in a 4.9\% improvement in success rate

Recent advances in natural language processing, primarily propelled by Large Language Models (LLMs), have showcased their remarkable capabilities grounded in in-context learning. A promising avenue for guiding LLMs in intricate reasoning tasks involves the utilization of intermediate reasoning steps within the Chain-of-Thought (CoT) paradigm. Nevertheless, the central challenge lies in the effective selection of exemplars for facilitating in-context learning. In this study, we introduce a framework that leverages Dual Queries and Low-rank approximation Re-ranking (DQ-LoRe) to automatically select exemplars for in-context learning. Dual Queries first query LLM to obtain LLM-generated knowledge such as CoT, then query the retriever to obtain the final exemplars via both question and the knowledge. Moreover, for the second query, LoRe employs dimensionality reduction techniques to refine exemplar selection, ensuring close alignment with the input question's knowledge. Through extensive experiments, we demonstrate that DQ-LoRe significantly outperforms prior state-of-the-art methods in the automatic selection of exemplars for GPT-4, enhancing performance from 92.5\% to 94.2\%. Our comprehensive analysis further reveals that DQ-LoRe consistently outperforms retrieval-based approaches in terms of both performance and adaptability, especially in scenarios characterized by distribution shifts. DQ-LoRe pushes the boundaries of in-context learning and opens up new avenues for addressing complex reasoning challenges.

In this paper, we propose to create animatable avatars for interacting hands with 3D Gaussian Splatting (GS) and single-image inputs. Existing GS-based methods designed for single subjects often yield unsatisfactory results due to limited input views, various hand poses, and occlusions. To address these challenges, we introduce a novel two-stage interaction-aware GS framework that exploits cross-subject hand priors and refines 3D Gaussians in interacting areas. Particularly, to handle hand variations, we disentangle the 3D presentation of hands into optimization-based identity maps and learning-based latent geometric features and neural texture maps. Learning-based features are captured by trained networks to provide reliable priors for poses, shapes, and textures, while optimization-based identity maps enable efficient one-shot fitting of out-of-distribution hands. Furthermore, we devise an interaction-aware attention module and a self-adaptive Gaussian refinement module. These modules enhance image rendering quality in areas with intra- and inter-hand interactions, overcoming the limitations of existing GS-based methods. Our proposed method is validated via extensive experiments on the large-scale InterHand2.6M dataset, and it significantly improves the state-of-the-art performance in image quality. Code and models will be released upon acceptance.

Formal verification (FV) has witnessed growing significance with current emerging program synthesis by the evolving large language models (LLMs). However, current formal verification mainly resorts to symbolic verifiers or hand-craft rules, resulting in limitations for extensive and flexible verification. On the other hand, formal languages for automated theorem proving, such as Isabelle, as another line of rigorous verification, are maintained with comprehensive rules and theorems. In this paper, we propose FVEL, an interactive Formal Verification Environment with LLMs. Specifically, FVEL transforms a given code to be verified into Isabelle, and then conducts verification via neural automated theorem proving with an LLM. The joined paradigm leverages the rigorous yet abundant formulated and organized rules in Isabelle and is also convenient for introducing and adjusting cutting-edge LLMs. To achieve this goal, we extract a large-scale FVELER. The FVELER dataset includes code dependencies and verification processes that are formulated in Isabelle, containing 758 theories, 29,304 lemmas, and 201,498 proof steps in total with in-depth dependencies. We benchmark FVELER in the FVEL environment by first fine-tuning LLMs with FVELER and then evaluating them on Code2Inv and SV-COMP. The results show that FVEL with FVELER fine-tuned Llama3-8B solves 17.39% (69→81) more problems, and Mistral-7B 12% (75→84) more problems in SV-COMP. And the proportion of proof errors is reduced. Project page: https://fveler.github.io/.

Recent advances in automated theorem proving leverages language models to explore expanded search spaces by step-by-step proof generation. However, such approaches are usually based on short-sighted heuristics (e.g., log probability or value function scores) that potentially lead to suboptimal or even distracting subgoals, preventing us from finding longer proofs. To address this challenge, we propose POETRY (PrOvE Theorems RecursivelY), which proves theorems in a recursive, level-by-level manner in the Isabelle theorem prover. Unlike previous step-by-step methods, POETRY searches for a verifiable sketch of the proof at each level and focuses on solving the current level's theorem or conjecture. Detailed proofs of intermediate conjectures within the sketch are temporarily replaced by a placeholder tactic called sorry, deferring their proofs to subsequent levels. This approach allows the theorem to be tackled incrementally by outlining the overall theorem at the first level and then solving the intermediate conjectures at deeper levels. Experiments are conducted on the miniF2F and PISA datasets and significant performance gains are observed in our POETRY approach over state-of-the-art methods. POETRY on miniF2F achieves an average proving success rate improvement of 5.1%. Moreover, we observe a substantial increase in the maximum proof length found by POETRY, from 10 to 26.

Recent advancements utilizing large-scale video data for learning video generation models demonstrate significant potential in understanding complex physical dynamics. It suggests the feasibility of leveraging diverse robot trajectory data to develop a unified, dynamics-aware model to enhance robot manipulation. However, given the relatively small amount of available robot data, directly fitting data without considering the relationship between visual observations and actions could lead to suboptimal data utilization. To this end, we propose \textbf{VidMan} (\textbf{Vid}eo Diffusion for Robot \textbf{Man}ipulation), a novel framework that employs a two-stage training mechanism inspired by dual-process theory from neuroscience to enhance stability and improve data utilization efficiency. Specifically, in the first stage, VidMan is pre-trained on the Open X-Embodiment dataset (OXE) for predicting future visual trajectories in a video denoising diffusion manner, enabling the model to develop a long horizontal awareness of the environment's dynamics. In the second stage, a flexible yet effective layer-wise self-attention adapter is introduced to transform VidMan into an efficient inverse dynamics model that predicts action modulated by the implicit dynamics knowledge via parameter sharing. Our VidMan framework outperforms state-of-the-art baseline model GR-1 on the CALVIN benchmark, achieving a 11.7\% relative improvement, and demonstrates over 9\% precision gains on the OXE small-scale dataset. These results provide compelling evidence that world models can significantly enhance the precision of robot action prediction. Codes and models will be public.

Multimodal large language models (MLLMs) have shown impressive success across modalities such as image, video, and audio in a variety of understanding and generation tasks. However, current MLLMs are surprisingly poor at understanding webpage screenshots and generating their corresponding HTML code. To address this problem, we propose Web2Code, a benchmark consisting of a new large-scale webpage-to-code dataset for instruction tuning and an evaluation framework for the webpage understanding and HTML code translation abilities of MLLMs. For dataset construction, we leverage pretrained LLMs to enhance existing webpage-to-code datasets as well as generate a diverse pool of new webpages rendered into images. Specifically, the inputs are webpage images and instructions, while the responses are the webpage's HTML code. We further include diverse natural language QA pairs about the webpage content in the responses to enable a more comprehensive understanding of the web content. To evaluate model performance in these tasks, we develop an evaluation framework for testing MLLMs' abilities in webpage understanding and web-to-code generation. Extensive experiments show that our proposed dataset is beneficial not only to our proposed tasks but also in the general visual domain. We hope our work will contribute to the development of general MLLMs suitable for web-based content generation and task automation. Our data and code are available at https://github.com/MBZUAI-LLM/web2code.

Language-guided robotic manipulation is a challenging task that requires an embodied agent to follow abstract user instructions to accomplish various complex manipulation tasks. Previous work generally maps instructions and visual perceptions directly to low-level executable actions, neglecting the modeling of critical waypoints (e.g., key states of “close to/grab/move up” in action trajectories) in manipulation tasks. To address this issue, we propose a PImitive-driVen waypOinT-aware world model for Robotic manipulation (PIVOT-R) that focuses solely on the prediction of task-relevant waypoints. Specifically, PIVOT-R consists of a Waypoint-aware World Model (WAWM) and a lightweight action prediction module. The former performs primitive action parsing and primitive-driven waypoint prediction, while the latter focuses on decoding low-level actions. Additionally, we also design an asynchronous hierarchical executor (AHE) for PIVOT-R, which can use different execution frequencies for different modules of the model, thereby helping the model reduce computational redundancy and improve model execution efficiency. Our PIVOT-R outperforms state-of-the-art (SoTA) open-source models on the SeaWave benchmark, achieving an average relative improvement of 19.45% across four levels of instruction tasks. Moreover, compared to the synchronously executed PIVOT-R, the execution efficiency of PIVOT-R with AHE is increased by 28-fold, with only a 2.9% drop in performance. These results provide compelling evidence that our PIVOT-R can significantly improve both the performance and efficiency of robotic manipulation.

Vision-Language Navigation (VLN) requires the agent to follow language instructions to reach a target position. A key factor for successful navigation is to align the landmarks implied in the instruction with diverse visual observations. However, previous VLN agents fail to perform accurate modality alignment especially in unexplored scenes, since they learn from limited navigation data and lack sufficient open-world alignment knowledge. In this work, we propose a new VLN paradigm, called COrrectable LaNdmark DiScOvery via Large ModEls (CONSOLE). In CONSOLE, we cast VLN as an open-world sequential landmark discovery problem, by introducing a novel correctable landmark discovery scheme based on two large models ChatGPT and CLIP. Specifically, we use ChatGPT to provide rich open-world landmark cooccurrence commonsense, and conduct CLIP-driven landmark discovery based on these commonsense priors. To mitigate the noise in the priors due to the lack of visual constraints, we introduce a learnable cooccurrence scoring module, which corrects the importance of each cooccurrence according to actual observations for accurate landmark discovery. We further design an observation enhancement strategy for an elegant combination of our framework with different VLN agents, where we utilize the corrected landmark features to obtain enhanced observation features for action decision. Extensive experimental results on multiple popular VLN benchmarks (R2R, REVERIE, R4R, RxR) show the significant superiority of CONSOLE over strong baselines. Especially, our CONSOLE establishes the new state-of-the-art results on R2R and R4R in unseen scenarios.

Neural Architecture Search (NAS), aiming at automatically designing neural architectures by machines, has been considered a key step toward automatic machine learning. One notable NAS branch is the weight-sharing NAS, which significantly improves search efficiency and allows NAS algorithms to run on ordinary computers. Despite receiving high expectations, this category of methods suffers from low search effectiveness. By employing a generalization boundedness tool, we demonstrate that the devil behind this drawback is the untrustworthy architecture rating with the oversized search space of the possible architectures. Addressing this problem, we modularize a large search space into blocks with small search spaces and develop a family of models with the distilling neural architecture (DNA) techniques. These proposed models, namely a DNA family, are capable of resolving multiple dilemmas of the weight-sharing NAS, such as scalability, efficiency, and multi-modal compatibility. Our proposed DNA models can rate all architecture candidates, as opposed to previous works that can only access a sub- search space using heuristic algorithms. Moreover, under a certain computational complexity constraint, our method can seek architectures with different depths and widths. Extensive experimental evaluations show that our models achieve state-of-the-art top-1 accuracy of 78.9% and 83.6% on ImageNet for a mobile convolutional network and a small vision transformer, respectively. Additionally, we provide in-depth empirical analysis and insights into neural architecture ratings.

Recently, there has been increasing interest in the challenge of how to discriminatively vectorize graphs. To address this, we propose a method called Iterative Graph Self-Distillation (IGSD) which learns graph-level representation in an unsupervised manner through instance discrimination using a self-supervised contrastive learning approach. IGSD involves a teacher-student distillation process that uses graph diffusion augmentations and constructs the teacher model using an exponential moving average of the student model. The intuition behind IGSD is to predict the teacher network representation of the graph pairs under different augmented views. As a natural extension, we also apply IGSD to semi-supervised scenarios by jointly regularizing the network with both supervised and self-supervised contrastive loss. Finally, we show that fine-tuning the IGSD-trained models with self-training can further improve graph representation learning. Empirically, we achieve significant and consistent performance gain on various graph datasets in both unsupervised and semi-supervised settings, which well validates the superiority of IGSD.

Graph-level representations are critical in various real-world applications, such as predicting the properties of molecules. However, in practice, precise graph annotations are generally very expensive and time-consuming. To address this issue, graph contrastive learning constructs an instance discrimination task, which pulls together positive pairs (augmentation pairs of the same graph) and pushes away negative pairs (augmentation pairs of different graphs) for unsupervised representation learning. However, since for a query, its negatives are uniformly sampled from all graphs, existing methods suffer from the critical sampling bias issue, i.e., the negatives likely having the same semantic structure with the query, leading to performance degradation. To mitigate this sampling bias issue, in this article, we propose a prototypical graph contrastive learning (PGCL) approach. Specifically, PGCL models the underlying semantic structure of the graph data via clustering semantically similar graphs into the same group and simultaneously encourages the clustering consistency for different augmentations of the same graph. Then, given a query, it performs negative sampling via drawing the graphs from those clusters that differ from the cluster of query, which ensures the semantic difference between query and its negative samples. Moreover, for a query, PGCL further reweights its negative samples based on the distance between their prototypes (cluster centroids) and the query prototype such that those negatives having moderate prototype distance enjoy relatively large weights. This reweighting strategy is proven to be more effective than uniform sampling. Experimental results on various graph benchmarks testify the advantages of our PGCL over state-of-the-art methods. The code is publicly available at https://github.com/ha-lins/PGCL.

Inspired by the success of vision-language methods (VLMs) in zero-shot classification, recent works attempt to extend this line of work into object detection by leveraging the localization ability of pretrained VLMs and generating pseudolabels for unseen classes in a self-training manner. However, since the current VLMs are usually pretrained with aligning sentence embedding with global image embedding, the direct use of them lacks fine-grained alignment for object instances, which is the core of detection. In this article, we propose a simple but effective fine-grained visual-text prompt-driven self-training paradigm for open-vocabulary detection (VTP-OVD) that introduces a fine-grained visual-text prompt adapting stage to enhance the current self-training paradigm with a more powerful fine-grained alignment. During the adapting stage, we enable VLM to obtain fine-grained alignment using learnable text prompts to resolve an auxiliary dense pixelwise prediction task. Furthermore, we propose a visual prompt module to provide the prior task information (i.e., the categories need to be predicted) for the vision branch to better adapt the pretrained VLM to the downstream tasks. Experiments show that our method achieves the state-of-the-art performance for open-vocabulary object detection, e.g., 31.5% mAP on unseen classes of COCO.

Since math word problem (MWP) solving aims to transform natural language problem description into executable solution equations, an MWP solver needs to not only comprehend the real-world narrative described in the problem text but also identify the relationships among the quantifiers and variables implied in the problem and maps them into a reasonable solution equation logic. Recently, although deep learning models have made great progress in MWPs, they ignore the grounding equation logic implied by the problem text. Besides, as we all know, pretrained language models (PLM) have a wealth of knowledge and high-quality semantic representations, which may help solve MWPs, but they have not been explored in the MWP-solving task. To harvest the equation logic and real-world knowledge, we propose a template-based contrastive distillation pretraining (TCDP) approach based on a PLM-based encoder to incorporate mathematical logic knowledge by multiview contrastive learning while retaining rich real-world knowledge and high-quality semantic representation via knowledge distillation. We named the pretrained PLM-based encoder by our approach as MathEncoder. Specifically, the mathematical logic is first summarized by clustering the symbolic solution templates among MWPs and then injected into the deployed PLM-based encoder by conducting supervised contrastive learning based on the symbolic solution templates, which can represent the underlying solving logic in the problems. Meanwhile, the rich knowledge and high-quality semantic representation are retained by distilling them from a well-trained PLM-based teacher encoder into our MathEncoder. To validate the effectiveness of our pretrained MathEncoder, we construct a new solver named MathSolver by replacing the GRU-based encoder with our pretrained MathEncoder in GTS, which is a state-of-the-art MWP solver. The experimental results demonstrate that our method can carry a solver's understanding ability of MWPs to a new stage by outperforming existing state-of-the-art methods on two widely adopted benchmarks Math23K and CM17K. Code will be available at https://github.com/QinJinghui/tcdp.

Large-scale cross-modal pre-training paradigms have recently shown ubiquitous success on a wide range of downstream tasks, e.g., zero-shot classification, retrieval and image captioning. However, their successes highly rely on the scale and quality of web-crawled data that naturally contain much incomplete and noisy information (e.g., wrong or irrelevant contents). Existing works either design manual rules to clean data or generate pseudo-targets as auxiliary signals for reducing noise impact, which do not explicitly tackle both the incorrect and incomplete challenges at the same time. In this paper, to automatically mitigate the impact of noise by solely mining over existing data, we propose a principled Noise-robust Language-Image Pre-training framework (NLIP) to stabilize pre-training via two schemes: noise-harmonization and noise-completion. First, in noise-harmonization scheme, NLIP estimates the noise probability of each pair according to the memorization effect of cross-modal transformers, then adopts noise-adaptive regularization to harmonize the cross-modal alignments with varying degrees. Second, in noise-completion scheme, to enrich the missing object information of text, NLIP injects a concept-conditioned cross-modal decoder to obtain semantic-consistent synthetic captions to complete noisy ones, which uses the retrieved visual concepts (i.e., objects’ names) for the corresponding image to guide captioning generation. By collaboratively optimizing noise-harmonization and noise-completion schemes, our NLIP can alleviate the common noise effects during image-text pre-training in a more efficient way. Extensive experiments show the significant performance improvements of our NLIP using only 26M data over existing pre-trained models (e.g., CLIP, FILIP and BLIP) on 12 zero-shot classification datasets (e.g., +8.6% over CLIP on average accuracy), MSCOCO image captioning (e.g., +1.9 over BLIP trained with 129M data on CIDEr) and zero-shot image-text retrieval tasks.

Text-guided 3D object generation aims to generate 3D objects described by user-defined captions, which paves a flexible way to visualize what we imagined. Although some works have been devoted to solving this challenging task, these works either utilize some explicit 3D representations (e.g., mesh), which lack texture and require post-processing for rendering photo-realistic views; or require individual time-consuming optimization for every single case. Here, we make the first attempt to achieve generic text-guided cross-category 3D object generation via a new 3D-TOGO model, which integrates a text-to-views generation module and a views-to-3D generation module. The text-to-views generation module is designed to generate different views of the target 3D object given an input caption. prior-guidance, caption-guidance and view contrastive learning are proposed for achieving better view-consistency and caption similarity. Meanwhile, a pixelNeRF model is adopted for the views-to-3D generation module to obtain the implicit 3D neural representation from the previously-generated views. Our 3D-TOGO model generates 3D objects in the form of the neural radiance field with good texture and requires no time-cost optimization for every single caption. Besides, 3D-TOGO can control the category, color and shape of generated 3D objects with the input caption. Extensive experiments on the largest 3D object dataset (i.e., ABO) are conducted to verify that 3D-TOGO can better generate high-quality 3D objects according to the input captions across 98 different categories, in terms of PSNR, SSIM, LPIPS and CLIP-score, compared with text-NeRF and Dreamfields.

Vision-Language Navigation (VLN) is a challenging task which requires an agent to align complex visual observations to language instructions to reach the goal position. Most existing VLN agents directly learn to align the raw directional features and visual features trained using one-hot labels to linguistic instruction features. However, the big semantic gap among these multi-modal inputs makes the alignment difficult and therefore limits the navigation performance. In this paper, we propose Actional Atomic-Concept Learning (AACL), which maps visual observations to actional atomic concepts for facilitating the alignment. Specifically, an actional atomic concept is a natural language phrase containing an atomic action and an object, e.g., ``go up stairs''. These actional atomic concepts, which serve as the bridge between observations and instructions, can effectively mitigate the semantic gap and simplify the alignment. AACL contains three core components: 1) a concept mapping module to map the observations to the actional atomic concept representations through the VLN environment and the recently proposed Contrastive Language-Image Pretraining (CLIP) model, 2) a concept refining adapter to encourage more instruction-oriented object concept extraction by re-ranking the predicted object concepts by CLIP, and 3) an observation co-embedding module which utilizes concept representations to regularize the observation representations. Our AACL establishes new state-of-the-art results on both fine-grained (R2R) and high-level (REVERIE and R2R-Last) VLN benchmarks. Moreover, the visualization shows that AACL significantly improves the interpretability in action decision. Code will be available at https://gitee.com/mindspore/models/tree/master/research/cv/VLN-AACL.

Haiming Wang, Ye Yuan, Zhengying Liu, Jianhao Shen, Yichun Yin, Jing Xiong, Enze Xie, Han Shi, Yujun Li, Lin Li, Jian Yin, Zhenguo Li, Xiaodan Liang. Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2023.

Automatic radiology reporting has great clinical potential to relieve radiologists from heavy workloads and improve diagnosis interpretation. Recently, researchers have enhanced data-driven neural networks with medical knowledge graphs to eliminate the severe visual and textual bias in this task. The structures of such graphs are exploited by using the clinical dependencies formed by the disease topic tags via general knowledge and usually do not update during the training process. Consequently, the fixed graphs can not guarantee the most appropriate scope of knowledge and limit the effectiveness. To address the limitation, we propose a knowledge graph with Dynamic structure and nodes to facilitate chest X-ray report generation with Contrastive Learning, named DCL. In detail, the fundamental structure of our graph is pre-constructed from general knowledge. Then we explore specific knowledge extracted from the retrieved reports to add additional nodes or redefine their relations in a bottom-up manner. Each image feature is integrated with its very own updated graph before being fed into the decoder module for report generation. Finally, this paper introduces Image-Report Contrastive and Image-Report Matching losses to better represent visual features and textual information. Evaluated on IU-Xray and MIMIC-CXR datasets, our DCL outperforms previous state-of-the-art models on these two benchmarks.

Multi-task learning has emerged as a powerful paradigm to solve a range of tasks simultaneously with good efficiency in both computation resources and inference time. However, these algorithms are designed for different tasks mostly not within the scope of autonomous driving, thus making it hard to compare multi-task methods in autonomous driving. Aiming to enable the comprehensive evaluation of present multi-task learning methods in autonomous driving, we extensively investigate the performance of popular multi-task methods on the large-scale driving dataset, which covers four common perception tasks, i.e., object detection, semantic segmentation, drivable area segmentation, and lane detection. We provide an in-depth analysis of current multi-task learning methods under different common settings and find out that the existing methods make progress but there is still a large performance gap compared with single-task baselines. To alleviate this dilemma in autonomous driving, we present an effective multi-task framework, VE-Prompt, which introduces visual exemplars via task-specific prompting to guide the model toward learning high-quality task-specific representations. Specifically, we generate visual exemplars based on bounding boxes and color-based markers, which provide accurate visual appearances of target categories and further mitigate the performance gap. Furthermore, we bridge transformer-based encoders and convolutional layers for efficient and accurate unified perception in autonomous driving. Comprehensive experimental results on the diverse self-driving dataset BDD100K show that the VE-Prompt improves the multi-task baseline and further surpasses single-task models.

Benefiting from large-scale vision-language pre-training on image-text pairs, open-world detection methods have shown superior generalization ability under the zero-shot or few-shot detection settings. However, a pre-defined category space is still required during the inference stage of existing methods and only the objects belonging to that space will be predicted. To introduce a “real” open-world detector, in this paper, we propose a novel method named CapDet to either predict under a given category list or directly generate the category of predicted bounding boxes. Specifically, we unify the open-world detection and dense caption tasks into a single yet effective framework by introducing an additional dense captioning head to generate the region-grounded captions. Besides, adding the captioning task will in turn benefit the generalization of detection performance since the captioning dataset covers more concepts. Experiment results show that by unifying the dense caption task, our CapDet has obtained significant performance improvements (e.g., +2.1% mAP on LVIS rare classes) over the baseline method on LVIS (1203 classes). Besides, our CapDet also achieves state-of-the-art performance on dense captioning tasks, e.g., 15.44% mAP on VG V1.2 and 13.98% on the VG-COCO dataset.

Referring Expression Segmentation (RES) can facilitate pixel-level semantic alignment between vision and language. Most of the existing RES approaches require massive pixel-level annotations, which are expensive and exhaustive. In this paper, we propose a new partially supervised training paradigm for RES, i.e., training using abundant referring bounding boxes and only a few (e.g., 1%) pixel-level referring masks. To maximize the transferability from the REC model, we construct our model based on the point-based sequence prediction model. We propose the co-content teacher-forcing to make the model explicitly associate the point coordinates (scale values) with the referred spatial features, which alleviates the exposure bias caused by the limited segmentation masks. To make the most of referring bounding box annotations, we further propose the resampling pseudo points strategy to select more accurate pseudo-points as supervision. Extensive experiments show that our model achieves 52.06% in terms of accuracy (versus 58.93% in fully supervised setting) on Re-fCOCO+@testA, when only using 1% of the mask annotations. Code is available at https://github.com/qumengxue/Partial-RES.git.

Image-based Virtual Try-ON aims to transfer an in-shop garment onto a specific person. Existing methods employ a global warping module to model the anisotropic deformation for different garment parts, which fails to preserve the semantic information of different parts when receiving challenging inputs (e.g, intricate human poses, difficult garments). Moreover, most of them directly warp the input garment to align with the boundary of the preserved region, which usually requires texture squeezing to meet the boundary shape constraint and thus leads to texture distortion. The above inferior performance hinders existing methods from real-world applications. To address these problems and take a step towards real-world virtual try-on, we propose a General-Purpose Virtual Try-ON framework, named GP-VTON, by developing an innovative Local-Flow Global-Parsing (LFGP) warping module and a Dynamic Gradient Truncation (DGT) training strategy. Specifically, compared with the previous global warping mechanism, LFGP employs local flows to warp garments parts individually, and assembles the local warped results via the global garment parsing, resulting in reasonable warped parts and a semantic-correct intact garment even with challenging inputs. On the other hand, our DGT training strategy dynamically truncates the gradient in the overlap area and the warped garment is no more required to meet the boundary constraint, which effectively avoids the texture squeezing problem. Furthermore, our GP-VTON can be easily extended to multi-category scenario and jointly trained by using data from different garment categories. Extensive experiments on two high-resolution benchmarks demonstrate our superiority over the existing state-of-the-art methods.11Code is available at gp-vton.

This paper presents DetCLIPv2, an efficient and scalable training framework that incorporates large-scale imagetext pairs to achieve open-vocabulary object detection (OVD). Unlike previous OVD frameworks that typically rely on a pre-trained vision-language model (e.g., CLIP) or exploit image-text pairs via a pseudo labeling process, DetCLIPv2 directly learns the fine-grained word-region alignment from massive image-text pairs in an end-to-end manner. To accomplish this, we employ a maximum word-region similarity between region proposals and textual words to guide the contrastive objective. To enable the model to gain localization capability while learning broad concepts, DetCLIPv2 is trained with a hybrid supervision from detection, grounding and image-text pair data under a unified data formulation. By jointly training with an alternating scheme and adopting low-resolution input for image-text pairs, DetCLIPv2 exploits image-text pair data efficiently and effectively: DetCLIPv2 utilizes 13 × more image-text pairs than DetCLIP with a similar training time and improves performance. With 13M image-text pairs for pre-training, DetCLIPv2 demonstrates superior open-vocabulary detection performance, e.g., DetCLIPv2 with Swin-T backbone achieves 40.4% zero-shot AP on the LVIS benchmark, which outperforms previous works GLIP/GLIPv2/DetCLIP by 14.4/11.4/4.5% AP, respectively, and even beats its fully-supervised counterpart by a large margin.

Contrastive Language-Image Pre-training, benefiting from large-scale unlabeled text-image pairs, has demonstrated great performance in open-world vision understanding tasks. However, due to the limited Text-3D data pairs, adapting the success of 2D Vision-Language Models (VLM) to the 3D space remains an open problem. Existing works that leverage VLM for 3D understanding generally resort to constructing intermediate 2D representations for the 3D data, but at the cost of losing 3D geometry information. To take a step toward open-world 3D vision understanding, we propose Contrastive Language-Image-Point Cloud Pretraining (CLIP2) to directly learn the transferable 3D point cloud representation in realistic scenarios with a novel proxy alignment mechanism. Specifically, we exploit naturally-existed correspondences in 2D and 3D scenarios, and build well-aligned and instance-based text-image-point proxies from those complex scenarios. On top of that, we propose a cross-modal contrastive objective to learn semantic and instance-level aligned point cloud representation. Experimental results on both indoor and outdoor scenarios show that our learned 3D representation has great transfer ability in downstream tasks, including zero-shot and few-shot 3D recognition, which boosts the state-of-the-art methods by large margins. Furthermore, we provide analyses of the capability of different representations in real scenarios and present the optional ensemble scheme.

Guangyi Liu, Zeyu Feng, Yuan Gao, Zichao Yang, Xiaodan Liang, Junwei Bao, Xiaodong He, Shuguang Cui, Zhen Li, Zhiting Hu. Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing. 2023.

Jing Xiong, Jianhao Shen, Ye Yuan, Haiming Wang, Yichun Yin, Zhengying Liu, Lin Li, Zhijiang Guo, Qingxing Cao, Yinya Huang, Chuanyang Zheng, Xiaodan Liang, Ming Zhang, Qun Liu. Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing. 2023.

Recently, semantic segmentation models trained with image-level text supervision have shown promising results in challenging open-world scenarios. However, these models still face difficulties in learning fine-grained semantic alignment at the pixel level and predicting accurate object masks. To address this issue, we propose MixReorg, a novel and straightforward pre-training paradigm for semantic segmentation that enhances a model’s ability to reorganize patches mixed across images, exploring both local visual relevance and global semantic coherence. Our approach involves generating fine-grained patch-text pairs data by mixing image patches while preserving the correspondence between patches and text. The model is then trained to minimize the segmentation loss of the mixed images and the two contrastive losses of the original and restored features. With MixReorg as a mask learner, conventional text-supervised semantic segmentation models can achieve highly generalizable pixel-semantic alignment ability, which is crucial for open-world segmentation. After training with large-scale image-text data, MixReorg models can be applied directly to segment visual objects of arbitrary categories, without the need for further fine-tuning. Our proposed framework demonstrates strong performance on popular zero-shot semantic segmentation benchmarks, outperforming GroupViT by significant margins of 5.0%, 6.2%, 2.5%, and 3.4% mIoU on PASCAL VOC2012, PASCAL Context, MS COCO, and ADE20K, respectively.

Cross-modal pre-training has shown impressive performance on a wide range of downstream tasks, benefiting from massive image-text pairs collected from the Internet. In practice, online data are growing constantly, highlighting the importance of the ability of pre-trained model to learn from data that is continuously growing. Existing works on cross-modal pre-training mainly focus on training a network with fixed architecture. However, it is impractical to limit the model capacity when considering the continuously growing nature of pre-training data in real-world applications. On the other hand, it is important to utilize the knowledge in the current model to obtain efficient training and better performance. To address the above issues, in this paper, we propose GrowCLIP, a data-driven automatic model growing algorithm for contrastive language-image pre-training with continuous image-text pairs as input. Specially, we adopt a dynamic growth space and seek out the optimal architecture at each growth step to adapt to online learning scenarios. And the shared encoder is proposed in our growth space to enhance the degree of cross-modal fusion. Besides, we explore the effect of growth in different dimensions, which could provide future references for the design of cross-modal model architecture. Finally, we employ parameter inheriting with momentum (PIM) to maintain the previous knowledge and address the issue of the local minimum dilemma. Compared with the existing methods, GrowCLIP improves 2.3% average top-1 accuracy on zero-shot image classification of 9 downstream tasks. As for zero-shot image retrieval, GrowCLIP can improve 1.2% for top-1 image-to-text recall on Flickr30K dataset.

Recently, large-scale diffusion models, e.g., Stable diffusion and DallE2, have shown remarkable results on image synthesis. On the other hand, large-scale cross-modal pre-trained models (e.g., CLIP, ALIGN, and FILIP) are competent for various downstream tasks by learning to align vision and language embeddings. In this paper, we explore the possibility of jointly modeling generation and discrimination. Specifically, we propose DiffDis to unify the cross-modal generative and discriminative pretraining into one single framework under the diffusion process. DiffDis first formulates the image-text discriminative problem as a generative diffusion process of the text embedding from the text encoder conditioned on the image. Then, we propose a novel dual-stream network architecture, which fuses the noisy text embedding with the knowledge of latent images from different scales for image-text discriminative learning. Moreover, the generative and discriminative tasks can efficiently share the image-branch network structure in the multi-modality model. Benefiting from diffusion-based unified training, DiffDis achieves both better generation ability and cross-modal semantic alignment in one architecture. Experimental results show that DiffDis outperforms single-task models on both the image generation and the image-text discriminative tasks, e.g., 1.65% improvement on average accuracy of zero-shot classification over 12 datasets and 2.42 improvement on FID of zero-shot image synthesis.

Multi-modality fusion and multi-task learning are becoming trendy in 3D autonomous driving scenario, considering robust prediction and computation budget. However, naively extending the existing framework to the domain of multi-modality multi-task learning remains ineffective and even poisonous due to the notorious modality bias and task conflict. Previous works manually coordinate the learning framework with empirical knowledge, which may lead to sub-optima. To mitigate the issue, we propose a novel yet simple multi-level gradient calibration learning framework across tasks and modalities during optimization. Specifically, the gradients, produced by the task heads and used to update the shared backbone, will be calibrated at the backbone’s last layer to alleviate the task conflict. Before the calibrated gradients are further propagated to the modality branches of the backbone, their magnitudes will be calibrated again to the same level, ensuring the downstream tasks pay balanced attention to different modalities. Experiments on large-scale benchmark nuScenes demonstrate the effectiveness of the proposed method, e.g., an absolute 14.4% mIoU improvement on map segmentation and 1.4% mAP improvement on 3D detection, advancing the application of 3D autonomous driving in the domain of multi-modality fusion and multi-task learning. We also discuss the links between modalities and tasks.

Multi-person 3D mesh recovery from videos is a critical first step towards automatic perception of group behavior in virtual reality, physical therapy and beyond. However, existing approaches rely on multi-stage paradigms, where the person detection and tracking stages are performed in a multi-person setting, while temporal dynamics are only modeled for one person at a time. Consequently, their performance is severely limited by the lack of inter-person interactions in the spatial-temporal mesh recovery, as well as by detection and tracking defects. To address these challenges, we propose the Coordinate transFormer (Coord-Former) that directly models multi-person spatial-temporal relations and simultaneously performs multi-mesh recovery in an end-to-end manner Instead of partitioning the feature map into coarse-scale patch-wise tokens, CoordFormer leverages a novel Coordinate-Aware Attention to preserve pixel-level spatial-temporal coordinate information. Additionally, we propose a simple, yet effective Body Center Attention mechanism to fuse position information. Extensive experiments on the 3DPW dataset demonstrate that CoordFormer significantly improves the state-of-the-art, outperforming the previously best results by 4.2%, 8.8% and 4.7% according to the MPJPE, PAMPJPE, and PVE metrics, respectively, while being 40% faster than recent video-based approaches. The released code can be found at https://github.com/Li-Hao-yuan/CoordFormer.

Thanks to the rapid development of diffusion models, unprecedented progress has been witnessed in image synthesis. Prior works mostly rely on pre-trained linguistic models, but a text is often too abstract to properly specify all the spatial properties of an image, e.g., the layout configuration of a scene, leading to the sub-optimal results of complex scene generation. In this paper, we achieve accurate complex scene generation by proposing a semantically controllable Layout-AWare diffusion model, termed LAW-Diffusion. Distinct from the previous Layout-to-Image generation (L2I) methods that primarily explore category-aware relationships, LAW-Diffusion introduces a spatial dependency parser to encode the location-aware semantic coherence across objects as a layout embedding and produces a scene with perceptually harmonious object styles and contextual relations. To be specific, we delicately instantiate each object’s regional semantics as an object region map and leverage a location-aware cross-object attention module to capture the spatial dependencies among those disentangled representations. We further propose an adaptive guidance schedule for our layout guidance to mitigate the trade-off between the regional semantic alignment and the texture fidelity of generated objects. Moreover, LAW-Diffusion allows for instance reconfiguration while maintaining the other regions in a synthesized image by introducing a layout-aware latent grafting mechanism to recompose its local regional semantics. To better verify the plausibility of generated scenes, we propose a new evaluation metric for the L2I task, dubbed Scene Relation Score (SRS) to measure how the images preserve the rational and harmonious relations among contextual objects. Comprehensive experiments on COCO-Stuff and Visual-Genome demonstrate that our LAW-Diffusion yields the state-of-the-art generative performance, especially with coherent object relations.

Generating 3D faces from textual descriptions has a multitude of applications, such as gaming, movie, and robotics. Recent progresses have demonstrated the success of unconditional 3D face generation and text-to-3D shape generation. However, due to the limited text-3D face data pairs, text-driven 3D face generation remains an open problem. In this paper, we propose a text-guided 3D faces generation method, refer as TG-3DFace, for generating realistic 3D faces using text guidance. Specifically, we adopt an unconditional 3D face generation framework and equip it with text conditions, which learns the text-guided 3D face generation with only text-2D face data. On top of that, we propose two text-to-face cross-modal alignment techniques, including the global contrastive learning and the fine-grained alignment module, to facilitate high semantic consistency between generated 3D faces and input texts. Besides, we present directional classifier guidance during the inference process, which encourages creativity for out-of-domain generations. Compared to the existing methods, TG-3DFace creates more realistic and aesthetically pleasing 3D faces, boosting 9% multi-view consistency (MVIC) over Latent3D. The rendered face images generated by TG-3DFace achieve higher FID and CLIP score than text-to-2D face/image generation models, demonstrating our superiority in generating realistic and semantic-consistent textures.

Cross-modal garment synthesis and manipulation will significantly benefit the way fashion designers generate garments and modify their designs via flexible linguistic interfaces. However, despite the significant progress that has been made in generic image synthesis using diffusion models, producing garment images with garment part level semantics that are well aligned with input text prompts and then flexibly manipulating the generated results still remains a problem. Current approaches follow the general text-to-image paradigm and mine cross-modal relations via simple cross-attention modules, neglecting the structural correspondence between visual and textual representations in the fashion design domain. In this work, we instead introduce DiffCloth, a diffusion-based pipeline for cross-modal garment synthesis and manipulation, which empowers diffusion models with flexible compositionality in the fashion domain by structurally aligning the cross-modal semantics. Specifically, we formulate the part-level cross-modal alignment as a bipartite matching problem between the linguistic Attribute-Phrases (AP) and the visual garment parts which are obtained via constituency parsing and semantic segmentation, respectively. To mitigate the issue of attribute confusion, we further propose a semantic-bundled cross-attention to preserve the spatial structure similarities between the attention maps of attribute adjectives and part nouns in each AP. Moreover, DiffCloth allows for manipulation of the generated results by simply replacing APs in the text prompts. The manipulation-irrelevant regions are recognized by blended masks obtained from the bundled attention maps of the APs and kept unchanged. Extensive experiments on the CM-Fashion benchmark demonstrate that DiffCloth both yields state-of-the-art garment synthesis results by leveraging the inherent structural information and supports flexible manipulation with region consistency.

Vision-Language Pretraining (VLP) has shown impressive results on diverse downstream tasks by offline training on large-scale datasets. Regarding the growing nature of real-world data, such an offline training paradigm on ever-expanding data is unsustainable, because models lack the continual learning ability to accumulate knowledge constantly. However, most continual learning studies are limited to uni-modal classification and existing multimodal datasets cannot simulate continual non-stationary data stream scenarios. To support the study of Vision-Language Continual Pretraining (VLCP), we first contribute a comprehensive and unified benchmark dataset P9D which contains over one million product image-text pairs from 9 industries. The data from each industry as an independent task supports continual learning and conforms to the real-world long-tail nature to simulate pretraining on web data. We comprehensively study the characteristics and challenges of VLCP, and propose a new algorithm: Compatible momentum contrast with Topology Preservation, dubbed CTP. The compatible momentum model absorbs the knowledge of the current and previous-task models to flexibly update the modal feature. Moreover, Topology Preservation transfers the knowledge of embedding across tasks while preserving the flexibility of feature adjustment. The experimental results demonstrate our method not only achieves superior performance compared with other baselines but also does not bring an expensive training burden. Dataset and codes are available at https://github.com/KevinLight831/CTP.

Vision-language navigation (VLN) requires an agent to perceive visual observation in a house scene and navigate step-by-step following natural language instruction. Due to the high cost of data annotation and data collection, current VLN datasets provide limited instruction-trajectory data samples. Learning vision-language alignment for VLN from limited data is challenging since visual observation and language instruction are both complex and diverse. Previous works only generate augmented data based on original scenes while failing to generate data samples from unseen scenes, which limits the generalization ability of the navigation agent. In this paper, we introduce the Knowledge-driven Environmental Dreamer (KED), a method that leverages the knowledge of the embodied environment and generates unseen scenes for a navigation agent to learn. Generating an unseen environment with texture consistency and structure consistency is challenging. To address this problem, we incorporate three knowledge-driven regularization objectives into the KED and adopt a reweighting mechanism for self-adaptive optimization. Our KED method is able to generate unseen embodied environments without extra annotations. We use KED to successfully generate 270 houses and 500K instruction-trajectory pairs. The navigation agent with the KED method outperforms the state-of-the-art methods on various VLN benchmarks, such as R2R, R4R, and RxR. Both qualitative and quantitative experiments prove that our proposed KED method is able to high-quality augmentation data with texture consistency and structure consistency.

Intention-oriented object detection aims to detect desired objects based on specific intentions or requirements. For instance, when we desire to "lie down and rest", we instinctively seek out a suitable option such as a "bed" or a "sofa" that can fulfill our needs. Previous work in this area is limited either by the number of intention descriptions or by the affordance vocabulary available for intention objects. These limitations make it challenging to handle intentions in open environments effectively. To facilitate this research, we construct a comprehensive dataset called Reasoning Intention-Oriented Objects (RIO). In particular, RIO is specifically designed to incorporate diverse real-world scenarios and a wide range of object categories. It offers the following key features: 1) intention descriptions in RIO are represented as natural sentences rather than a mere word or verb phrase, making them more practical and meaningful; 2) the intention descriptions are contextually relevant to the scene, enabling a broader range of potential functionalities associated with the objects; 3) the dataset comprises a total of 40,214 images and 130,585 intention-object pairs. With the proposed RIO, we evaluate the ability of some existing models to reason intention-oriented objects in open environments.

A significant challenge facing researchers in the area of multi-agent reinforcement learning (MARL) pertains to the identification of a library that can offer fast and compatible development for multi-agent tasks and algorithm combinations, while obviating the need to consider compatibility issues. In this paper, we present MARLlib, a library designed to address the aforementioned challenge by leveraging three key mechanisms: 1) a standardized multi-agent environment wrapper, 2) an agent-level algorithm implementation, and 3) a flexible policy mapping strategy. By utilizing these mechanisms, MARLlib can effectively disentangle the intertwined nature of the multi-agent task and the learning process of the algorithm, with the ability to automatically alter the training strategy based on the current task's attributes. The MARLlib library's source code is publicly accessible on GitHub: https://github.com/Replicable-MARL/MARLlib.

The success of deep learning has brought breakthroughs in many fields. However, the increased performance of deep learning models is often accompanied by an increase in their depth and width, which conflicts with the storage, energy consumption, and computational power of edge devices. Knowledge distillation, as an effective model compression method, can transfer knowledge from complex teacher models to student models. Self-distillation is a special type of knowledge distillation, which does not to require a pre-trained teacher model. However, existing self-distillation methods rarely consider how to effectively use the early features of the model. Furthermore, most self-distillation methods use features from the deepest layers of the network to guide the training of the branches of the network, which we find is not the optimal choice. In this paper, we found that the feature maps obtained by early feature fusion do not serve as a good teacher to guide their own training. Based on this, we propose a selective feature fusion module and further obtain a new self-distillation method, knowledge fusion distillation. Extensive experiments on three datasets have demonstrated that our method has comparable performance to state-of-the-art distillation methods. In addition, the performance of the network can be further enhanced when fused features are integrated into the network.

Our goal in this research is to study a more realistic environment in which we can conduct weakly-supervised multi-modal instance-level product retrieval for fine-grained product categories. We first contribute the Product1M datasets and define two real practical instance-level retrieval tasks that enable evaluations on price comparison and personalized recommendations. For both instance-level tasks, accurately identifying the intended product target mentioned in visual-linguistic data and mitigating the impact of irrelevant content are quite challenging. To address this, we devise a more effective cross-modal pretraining model capable of adaptively incorporating key concept information from multi-modal data. This is accomplished by utilizing an entity graph, where nodes represented entities and edges denoted the similarity relations between them. Specifically, a novel Entity-Graph Enhanced Cross-Modal Pretraining (EGE-CMP) model is proposed for instance-level commodity retrieval, which explicitly injects entity knowledge in both node-based and subgraph-based ways into the multi-modal networks via a self-supervised hybrid-stream transformer. This could reduce the confusion between different object contents, thereby effectively guiding the network to focus on entities with real semantics. Experimental results sufficiently verify the efficacy and generalizability of our EGE-CMP, outperforming several SOTA cross-modal baselines like CLIP Radford et al. 2021, UNITER Chen et al. 2020 and CAPTURE Zhan et al. 2021.

Textual logical reasoning, especially question-answering (QA) tasks with logical reasoning, requires awareness of particular logical structures. The passage-level logical relations represent entailment or contradiction between propositional units (e.g., a concluding sentence). However, such structures are unexplored as current QA systems focus on entity-based relations. In this work, we propose logic structural-constraint modeling to solve the logical reasoning QA and introduce discourse-aware graph networks (DAGNs). The networks first construct logic graphs leveraging in-line discourse connectives and generic logic theories, then learn logic representations by end-to-end evolving the logic relations with an edge-reasoning mechanism and updating the graph features. This pipeline is applied to a general encoder, whose fundamental features are joined with the high-level logic features for answer prediction. Experiments on three textual logical reasoning datasets demonstrate the reasonability of the logical structures built in DAGNs and the effectiveness of the learned logic features. Moreover, zero-shot transfer results show the features' generality to unseen logical texts.

Dynamic networks have shown their promising capability in reducing theoretical computation complexity by adapting their architectures to the input during inference. However, their practical runtime usually lags behind the theoretical acceleration due to inefficient sparsity. In this paper, we explore a hardware-efficient dynamic inference regime, named dynamic weight slicing, that can generalized well on multiple dimensions in both CNNs and transformers (e.g. kernel size, embedding dimension, number of heads, etc.). Instead of adaptively selecting important weight elements in a sparse way, we pre-define dense weight slices with different importance level by nested residual learning. During inference, weights are progressively sliced beginning with the most important elements to less important ones to achieve different model capacity for inputs with diverse difficulty levels. Based on this conception, we present DS-CNN++ and DS-ViT++, by carefully designing the double headed dynamic gate and the overall network architecture. We further propose dynamic idle slicing to address the drastic reduction of embedding dimension in DS-ViT++. To ensure sub-network generality and routing fairness, we propose a disentangled two-stage optimization scheme. In Stage I, in-place bootstrapping (IB) and multi-view consistency (MvCo) are proposed to stablize and improve the training of DS-CNN++ and DS-ViT++ supernet, respectively. In Stage II, sandwich gate sparsification (SGS) is proposed to assist the gate training. Extensive experiments on 4 datasets and 3 different network architectures demonstrate our methods consistently outperform the state-of-the-art static and dynamic model compression methods by a large margin (up to 6.6%). Typically, we achieves 2-4× computation reduction and up to 61.5% real-world acceleration on MobileNet, ResNet-50 and Vision Transformer, with minimal accuracy drops on ImageNet. Code release: https://github.com/changlin31/DS-Net.

Vision-and-language navigation (VLN) asks an agent to follow a given language instruction to navigate through a real 3D environment. Despite significant advances, conventional VLN agents are trained typically under disturbance-free environments and may easily fail in real-world navigation scenarios, since they are unaware of how to deal with various possible disturbances, such as sudden obstacles or human interruptions, which widely exist and may usually cause an unexpected route deviation. In this paper, we present a model-agnostic training paradigm, called Progressive Perturbation-aware Contrastive Learning (PROPER) to enhance the generalization ability of existing VLN agents to the real world, by requiring them to learn towards deviation-robust navigation. Specifically, a simple yet effective path perturbation scheme is introduced to implement the route deviation, with which the agent is required to still navigate successfully following the original instruction. Since directly enforcing the agent to learn perturbed trajectories may lead to insufficient and inefficient training, a progressively perturbed trajectory augmentation strategy is designed, where the agent can self-adaptively learn to navigate under perturbation with the improvement of its navigation performance for each specific trajectory. For encouraging the agent to well capture the difference brought by perturbation and adapt to both perturbation-free and perturbation-based environments, a perturbation-aware contrastive learning mechanism is further developed by contrasting perturbation-free trajectory encodings and perturbation-based counterparts. Extensive experiments on the standard Room-to-Room (R2R) benchmark show that PROPER can benefit multiple state-of-the-art VLN baselines in perturbation-free scenarios. We further collect the perturbed path data to construct an introspection subset based on the R2R, called Path-Perturbed R2R (PP-R2R). The results on PP-R2R show unsatisfying robustness of popular VLN agents and the capability of PROPER in improving the navigation robustness under deviation.

Medical diagnosis assistant (MDA) aims to build an interactive diagnostic agent to sequentially inquire about symptoms for discriminating diseases. However, since the dialogue records for building a patient simulator are collected passively, the collected records might be deteriorated by some task-unrelated biases, such as the preference of the collectors. These biases might hinder the diagnostic agent to capture transportable knowledge from the simulator. This work identifies and resolves two representative non-causal biases, i.e., (i) default-answer bias and (ii) distributional inquiry bias. Specifically, Bias (i) originates from the patient simulator which tries to answer the unrecorded inquiries with some biased default answers. To eliminate this bias and improve upon a well-known causal inference technique, i.e., propensity score matching, we propose a novel propensity latent matching in building a patient simulator to effectively answer unrecorded inquiries; Bias (ii) inherently comes along with the passively collected data that the agent might learn by remembering what to inquire within the training data while not able to generalize to the out-of-distribution cases. To this end, we propose a progressive assurance agent, which includes the dual processes accounting for symptom inquiry and disease diagnosis respectively. The diagnosis process pictures the patient mentally and probabilistically by intervention to eliminate the effect of the inquiry behavior. And the inquiry process is driven by the diagnosis process to inquire about symptoms to enhance the diagnostic confidence which alters as the patient distribution changes. In this cooperative manner, our proposed agent can improve upon the out-of-distribution generalization significantly. Extensive experiments demonstrate that our framework achieves new state-of-the-art performance and possesses the advantage of transportability.

Few-shot class-incremental learning (FSCIL) is challenged by catastrophically forgetting old classes and over-fitting new classes. Revealed by our analyses, the problems are caused by feature distribution crumbling, which leads to class confusion when continuously embedding few samples to a fixed feature space. In this study, we propose a Dynamic Support Network (DSN), which refers to an adaptively updating network with compressive node expansion to "support" the feature space. In each training session, DSN tentatively expands network nodes to enlarge feature representation capacity for incremental classes. It then dynamically compresses the expanded network by node self-activation to pursue compact feature representation, which alleviates over-fitting. Simultaneously, DSN selectively recalls old class distributions during incremental learning to support feature distributions and avoid confusion between classes. DSN with compressive node expansion and class distribution recalling provides a systematic solution for the problems of catastrophic forgetting and overfitting. Experiments on CUB, CIFAR-100, and miniImage datasets show that DSN significantly improves upon the baseline approach, achieving new state-of-the-arts.

3-D object detection is a fundamental task in the context of autonomous driving. In the literature, cheap monocular image-based methods show a significant performance drop compared to the expensive LiDAR and stereo-images-based algorithms. In this article, we aim to close this performance gap by bridging the representation capability between 2-D and 3-D domains. We propose a novel monocular 3-D object detection model using self-supervised learning and auxiliary learning, resorting to mimicking the representations over 3-D point clouds. Specifically, given a 2-D region proposal and the corresponding instance point cloud, we supervise the feature activation from our image-based convolution network to mimic the latent feature of a point-based neural network at the training stage. While state-of-the-art (SOTA) monocular 3-D detection algorithms typically convert images to pseudo-LiDAR with depth estimation and regress 3-D detection with LiDAR-based methods, our approach seeks the power of the 2-D neural network straightforwardly and essentially enhances the 2-D module capability with latent spatial-aware representations by contrastive learning. We empirically validate the performance improvement from the feature mimicking the KITTI and ApolloScape datasets and achieve the SOTA performance on the KITTI and ApolloScape leaderboard.

Product detection, which aims to localize products of interest in the advertising images, helps advance many potential E-commerce applications like product retrieval and recommendation. However, labeling a massive number of fine-grained product categories and accurate product boxes is costly and especially not practical since products are ever-changing on E-commerce websites. In this work, we step forward to train a fine-grained product detector solely supervised by the advertising captions, which are naturally available but often severely flawed and noisy. To reformulate the weakly supervised detection research into a real-world setting, we introduce a large-scale benchmark, namedCapProduct, where more than 80,000 product image-caption pairs are collected from E-commerce websites. The fine-grained nature of products and noisy captions inCapProductmake it intractable to excavate valid category labels to train a weakly supervised object detector. To tackle this challenge, we propose aCollaborativePseudo-LabelHarmonization (CoPLH) framework that harmonizes self-mined pseudo labels via modeling the global co-occurrence relationships of products. We construct a collaborative co-occurrence graph based on all training samples to improve the reliability of caption-predicted pseudo-labels as well as benefit the self-training procedure in a weakly supervised setting. Extensive experiments on theCapProductdataset demonstrate the effectiveness and the superiority of the proposed CoPLH over the state-of-the-art baselines.

Medical reports have significant clinical value to radiologists and specialists, especially during a pandemic like COVID. However, beyond the common difficulties faced in the natural image captioning, medical report generation specifically requires the model to describe a medical image with a fine-grained and semantic-coherence paragraph that should satisfy both medical commonsense and logic. Previous works generally extract the global image features and attempt to generate a paragraph that is similar to referenced reports; however, this approach has two limitations. Firstly, the regions of primary interest to radiologists are usually located in a small area of the global image, meaning that the remainder parts of the image could be considered as irrelevant noise in the training procedure. Secondly, there are many similar sentences used in each medical report to describe the normal regions of the image, which causes serious data bias. This deviation is likely to teach models to generate these inessential sentences on a regular basis. To address these problems, we propose an Auxiliary Signal-Guided Knowledge Encoder-Decoder (ASGK) to mimic radiologists' working patterns. Specifically, the auxiliary patches are explored to expand the widely used visual patch features before fed to the Transformer encoder, while the external linguistic signals help the decoder better master prior knowledge during the pre-training process. Our approach performs well on common benchmarks, including CX-CHR, IU X-Ray, and COVID-19 CT Report dataset (COV-CTR), demonstrating combining auxiliary signals with transformer architecture can bring a significant improvement in terms of medical report generation. The experimental results confirm that auxiliary signals driven Transformer-based models are with solid capabilities to outperform previous approaches on both medical terminology classification and paragraph generation metrics.

Transformer-based pre-trained language models like BERT and its variants have recently achieved promising performance in various natural language processing (NLP) tasks. However, the conventional paradigm constructs the backbone by purely stacking the manually designed global self-attention layers, introducing inductive bias and thus leads to sub-optimal. In this work, we make the first attempt to automatically discover novel pre-trained language model (PLM) backbone on a flexible search space containing the most fundamental operations from scratch. Specifically, we propose a well-designed search space which (i) contains primitive math operations in the intra-layer level to explore novel attention structures, and (ii) leverages convolution blocks to be the supplementary for attentions in the inter-layer level to better learn local dependency. To enhance the efficiency for finding promising architectures, we propose an Operation-Priority Neural Architecture Search (OP-NAS) algorithm, which optimizes both the search algorithm and evaluation of candidate models. Specifically, we propose Operation-Priority (OP) evolution strategy to facilitate model search via balancing exploration and exploitation. Furthermore, we design a Bi-branch Weight-Sharing (BIWS) training strategy for fast model evaluation. Extensive experiments show that the searched architecture (named AutoBERT-Zero) significantly outperforms BERT and its variants of different model capacities in various downstream tasks, proving the architecture's transfer and scaling abilities. Remarkably, AutoBERT-Zero-base outperforms RoBERTa-base (using much more data) and BERT-large (with much larger model size) by 2.4 and 1.4 higher score on GLUE test set.

We present Laneformer, a conceptually simple yet powerful transformer-based architecture tailored for lane detection that is a long-standing research topic for visual perception in autonomous driving. The dominant paradigms rely on purely CNN-based architectures which often fail in incorporating relations of long-range lane points and global contexts induced by surrounding objects (e.g., pedestrians, vehicles). Inspired by recent advances of the transformer encoder-decoder architecture in various vision tasks, we move forwards to design a new end-to-end Laneformer architecture that revolutionizes the conventional transformers into better capturing the shape and semantic characteristics of lanes, with minimal overhead in latency. First, coupling with deformable pixel-wise self-attention in the encoder, Laneformer presents two new row and column self-attention operations to efficiently mine point context along with the lane shapes. Second, motivated by the appearing objects would affect the decision of predicting lane segments, Laneformer further includes the detected object instances as extra inputs of multi-head attention blocks in the encoder and decoder to facilitate the lane point detection by sensing semantic contexts. Specifically, the bounding box locations of objects are added into Key module to provide interaction with each pixel and query while the ROI-aligned features are inserted into Value module. Extensive experiments demonstrate our Laneformer achieves state-of-the-art performances on CULane benchmark, in terms of 77.1% F1 score. We hope our simple and effective Laneformer will serve as a strong baseline for future research in self-attention models for lane detection.

The vision-language navigation (VLN) task requires an agent to reach a target with the guidance of natural language instruction. Previous works learn to navigate step-by-step following an instruction. However, these works may fail to discriminate the similarities and discrepancies across instruction-trajectory pairs and ignore the temporal continuity of sub-instructions. These problems hinder agents from learning distinctive vision-and-language representations, harming the robustness and generalizability of the navigation policy. In this paper, we propose a Contrastive Instruction-Trajectory Learning (CITL) framework that explores invariance across similar data samples and variance across different ones to learn distinctive representations for robust navigation. Specifically, we propose: (1) a coarse-grained contrastive learning objective to enhance vision-and-language representations by contrasting semantics of full trajectory observations and instructions, respectively; (2) a fine-grained contrastive learning objective to perceive instructions by leveraging the temporal information of the sub-instructions; (3) a pairwise sample-reweighting mechanism for contrastive learning to mine hard samples and hence mitigate the influence of data sampling bias in contrastive learning. Our CITL can be easily integrated with VLN backbones to form a new learning paradigm and achieve better generalizability in unseen environments. Extensive experiments show that the model with CITL surpasses the previous state-of-the-art methods on R2R, R4R, and RxR.

Vision-language navigation (VLN) is a challenging task due to its large searching space in the environment.To address this problem, previous works have proposed some methods of fine-tuning a large model that pretrained on large-scale datasets.However, the conventional fine-tuning methods require extra human-labeled navigation data and lack self-exploration capabilities in environments, which hinders their generalization of unseen scenes.To improve the ability of fast cross-domain adaptation, we propose Prompt-based Environmental Self-exploration (ProbES), which can self-explore the environments by sampling trajectories and automatically generates structured instructions via a large-scale cross-modal pretrained model (CLIP).Our method fully utilizes the knowledge learned from CLIP to build an in-domain dataset by self-exploration without human labeling.Unlike the conventional approach of fine-tuning, we introduce prompt-based learning to achieve fast adaptation for language embeddings, which substantially improves the learning efficiency by leveraging prior knowledge.By automatically synthesizing trajectoryinstruction pairs in any environment without human supervision and efficient prompt-based learning, our model can adapt to diverse visionlanguage navigation tasks, including VLN and REVERIE.Both qualitative and quantitative results show that our ProbES significantly improves the generalization ability of the navigation model * .

Despite the potential of multi-modal pre-training to learn highly discriminative feature representations from complementary data modalities, current progress is being slowed by the lack of large-scale modality-diverse datasets. By leveraging the natural suitability of E-commerce, where different modalities capture complementary semantic information, we contribute a large-scale multi-modal pretraining dataset M5Product. The dataset comprises 5 modalities (image, text, table, video, and audio), covers over 6,000 categories and 5,000 attributes, and is 500× larger than the largest publicly available dataset with a similar number of modalities. Furthermore, M5Product contains incomplete modality pairs and noise while also having a long-tailed distribution, resembling most real-world problems. We further propose Self-harmonized ContrAstive LEarning (SCALE), a novel pretraining framework that integrates the different modalities into a unified model through an adaptive feature fusion mechanism, where the importance of each modality is learned directly from the modality embeddings and impacts the inter-modality contrastive learning and masked tasks within a multi-modal transformer model. We evaluate the current multi-modal pre-training state-of-the-art approaches and benchmark their ability to learn from unlabeled data when faced with the large number of modalities in the M5Product dataset. We conduct extensive experiments on four downstream tasks and demonstrate the superiority of our SCALE model, providing insights into the importance of dataset scale and diversity. Dataset and codes are available at11https://xiaodongsuper.github.io/M5Product_dataset/.

While significant progress has been made in garment transfer, one of the most applicable directions of human-centric image generation, existing works overlook the in-the-wild imagery, presenting severe garment-person mis-alignment as well as noticeable degradation in fine texture details. This paper, therefore, attends to virtual try-on in real-world scenes and brings essential improvements in authenticity and naturalness especially for loose garment (e.g., skirts, formal dresses), challenging poses (e.g., cross arms, bent legs), and cluttered backgrounds. Specifically, we find that the pixel flow excels at handling loose gar-ments whereas the vertex flow is preferred for hard poses, and by combining their advantages we propose a novel generative network called wFlow that can effectively push up garment transfer to in-the-wild context. Moreover, former approaches require paired images for training. Instead, we cut down the laboriousness by working on a newly constructed large-scale video dataset named Dance50k with self-supervised cross-frame training and an online cycle op-timization. The proposed Dance50k can boost real-world virtual dressing by covering a wide variety of garments under dancing poses. Extensive experiments demonstrate the superiority of our w Flow in generating realistic garment transfer results for in-the-wild images without resorting to expensive paired datasets.11Xiaodan Liang is the corresponding author. The project page of wFlow is https://awesome-wflow.github.io.

Neural Architecture Search (NAS) aims to find efficient models for multiple tasks. Beyond seeking solutions for a single task, there are surging interests in transferring network design knowledge across multiple tasks. In this line of research, effectively modeling task correlations is vital yet highly neglected. Therefore, we propose Arch-Graph, a transferable NAS method that predicts task-specific optimal architectures with respect to given task embeddings. It leverages correlations across multiple tasks by using their embeddings as a part of the predictor's input for fast adaptation. We also formulate NAS as an architecture relation graph prediction problem, with the relational graph constructed by treating candidate architectures as nodes and their pairwise relations as edges. To enforce some basic properties such as acyclicity in the relational graph, we add additional constraints to the optimization process, converting NAS into the problem of finding a Maximal Weighted Acyclic Subgraph (MWAS). Our algorithm then strives to eliminate cycles and only establish edges in the graph if the rank results can be trusted. Through MWAS, Arch-Graph can effectively rank candidate models for each task with only a small budget to finetune the predictor. With extensive experiments on TransNAS-Bench-101, we show Arch-Graph's transferability and high sample efficiency across numerous tasks, beating many NAS methods designed for both single-task and multi-task search. It is able to find top 0.16% and 0.29% architectures on average on two search spaces under the budget of only 50 models.11Code: https://github.com/Centaurus982034/Arch-Graph

Automatic generation of ophthalmic reports using datadriven neural networks has great potential in clinical practice. When writing a report, ophthalmologists make inferences with prior clinical knowledge. This knowledge has been neglected in prior medical report generation methods. To endow models with the capability of incorporating expert knowledge, we propose a Cross-modal clinical Graph Transformer (CGT) for ophthalmic report generation (ORG), in which clinical relation triples are injected into the visual features as prior knowledge to drive the decoding procedure. However, two major common Knowledge Noise (KN) issues may affect models' effectiveness. 1) Existing general biomedical knowledge bases such as the UMLS may not align meaningfully to the specific context and language of the report, limiting their utility for knowledge injection. 2) Incorporating too much knowledge may divert the visual features from their correct meaning. To overcome these limitations, we design an automatic information extraction scheme based on natural language processing to obtain clinical entities and relations directly from in-domain training reports. Given a set of ophthalmic images, our CGT first restores a sub-graph from the clinical graph and injects the restored triples into visual features. Then visible matrix is employed during the encoding procedure to limit the impact of knowledge. Finally, reports are predicted by the encoded cross-modal features via a Transformer decoder. Extensive experiments on the large-scale FFA-IR benchmark demonstrate that the proposed CGT is able to outperform previous benchmark methods and achieve state-of-the-art performances.

Recent advances in vision Transformers (ViTs) have come with a voracious appetite for computing power, highlighting the urgent need to develop efficient training methods for ViTs. Progressive learning, a training scheme where the model capacity grows progressively during training, has started showing its ability in efficient training. In this paper, we take a practical step towards efficient training of ViTs by customizing and automating progressive learning. First, we develop a strong manual baseline for progressive learning of ViTs, by introducing momentum growth (MoGrow) to bridge the gap brought by model growth. Then, we propose automated progressive learning (AutoProg), an efficient training scheme that aims to achieve lossless acceleration by automatically increasing the training overload on-the-fly; this is achieved by adaptively deciding whether, where and how much should the model grow during progressive learning. Specifically, we first relax the optimization of the growth schedule to sub-network architecture optimization problem, then propose one-shot estimation of the sub-network performance via an elastic supernet. The searching overhead is reduced to minimal by recycling the parameters of the supernet. Extensive experiments of efficient training on ImageNet with two representative ViT models, DeiT and VOLO, demonstrate that AutoProg can accelerate ViTs training by up to 85.1% with no performance drop.11Code:https://github.com/changlin31/AutoProg.

Vision-Language Navigation (VLN) is a challenging task that requires an embodied agent to perform action-level modality alignment, i.e., make instruction-asked actions sequentially in complex visual environments. Most existing VLN agents learn the instruction-path data directly and cannot sufficiently explore action-level alignment knowledge inside the multi-modal inputs. In this paper, we propose modAlity-aligneD Action PrompTs (ADAPT), which provides the VLN agent with action prompts to enable the explicit learning of action-level modality alignment to pursue successful navigation. Specifically, an action prompt is defined as a modality-aligned pair of an image sub-prompt and a text sub-prompt, where the former is a single-view observation and the latter is a phrase like “walk past the chair”. When starting navigation, the instruction-related action prompt set is retrieved from a prebuilt action prompt base and passed through a prompt encoder to obtain the prompt feature. Then the prompt feature is concatenated with the original instruction feature and fed to a multilayer transformer for action prediction. To collect high-quality action prompts into the prompt base, we use the Contrastive Language-Image Pretraining (CLIP) model which has powerful cross-modality alignment ability. A modality alignment loss and a sequential consistency loss are further introduced to enhance the alignment of the action prompt and enforce the agent to focus on the related prompt sequentially. Experimental results on both R2R and RxR show the superiority of ADAPT over state-of-the-art methods.

Knowledge distillation becomes a de facto standard to improve the performance of small neural networks. Most of the previous works propose to regress the representational features from the teacher to the student in a one-to-one spatial matching fashion. However, people tend to overlook the fact that, due to the architecture differences, the semantic information on the same spatial location usually vary. This greatly undermines the underlying assumption of the one-to-one distillation approach. To this end, we propose a novel one-to-all spatial matching knowledge distillation approach. Specifically, we allow each pixel of the teacher feature to be distilled to all spatial locations of the student features given its similarity, which is generated from a target-aware transformer. Our approach surpasses the state-of-the-art methods by a significant margin on various computer vision benchmarks, such as ImageNet, Pascal VOC and COCOStuff10k. Code is available at https://github.com/sihaoevery/TaT.

Recently, a surge of interest in visual transformers is to reduce the computational cost by limiting the calculation of self-attention to a local window. Most current work uses a fixed single-scale window for modeling by default, ignoring the impact of window size on model performance. How-ever, this may limit the modeling potential of these window-based models for multi-scale information. In this paper, we propose a novel method, named Dynamic Window Vision Transformer (DW-ViT). The dynamic window strategy proposed by DW- ViT goes beyond the model that employs a fixed single window setting. To the best of our knowl-edge, we are the first to use dynamic multi-scale windows to explore the upper limit of the effect of window settings on model performance. In DW- ViT, multi-scale information is obtained by assigning windows of different sizes to different head groups of window multi-head self-attention. Then, the information is dynamically fused by assigning different weights to the multi-scale window branches. We con-ducted a detailed performance evaluation on three datasets, ImageNet-1K, ADE20K, and COCO. Compared with re-lated state-of-the-art (SoTA) methods, DW- ViT obtains the best performance. Specifically, compared with the current SoTA Swin Transformers [31], DW-ViT has achieved con-sistent and substantial improvements on all three datasets with similar parameters and computational costs. In addition, DW-ViT exhibits good scalability and can be easily inserted into any window-based visual transformers.11Code release: https://github.com/pzhren/DW-ViT. This work was done when the first author interned at Dark Matter AI..

Continual learning is a challenging real-world problem for constructing a mature AI system when data are provided in a streaming fashion. Despite recent progress in continual classification, the researches of continual object detection are impeded by the diverse sizes and numbers of objects in each image. Different from previous works that tune the whole network for all tasks, in this work, we present a simple and flexible framework for continual object detection via pRotOtypical taSk corrElaTion guided gaTing mechAnism (ROSETTA). Concretely, a unified framework is shared by all tasks while task-aware gates are introduced to automatically select sub-models for specific tasks. In this way, various knowledge can be successively memorized by storing their corresponding sub-model weights in this system. To make ROSETTA automatically determine which experience is available and useful, a prototypical task correlation guided Gating Diversity Controller (GDC) is introduced to adaptively adjust the diversity of gates for the new task based on class-specific prototypes. GDC module computes class-to-class correlation matrix to depict the cross-task correlation, and hereby activates more exclusive gates for the new task if a significant domain gap is observed. Comprehensive experiments on COCO-VOC, KITTI-Kitchen, class-incremental detection on VOC and sequential learning of four tasks show that ROSETTA yields state-of-the-art performance on both task-based and class-based continual object detection.11Codes are available at: https://github.com/dkxocl/ROSSETA.

Recent advances in generative adversarial networks (GANs) have provided potential solutions for photo-realistic human image synthesis. However, the explicit and individual control of synthesis over multiple factors, such as poses, body shapes, and skin colors, remains difficult for existing methods. This is because current methods mainly rely on a single pose/appearance model, which is limited in dis-entangling various poses and appearance in human images. In addition, such a unimodal strategy is prone to causing severe artifacts in the generated images like color distortions and unrealistic textures. To tackle these issues, this paper proposes a multi-factor conditioned method dubbed BodyGAN. Specifically, given a source image, our Body-GAN aims at capturing the characteristics of the human body from multiple aspects: (i) A pose encoding branch consisting of three hybrid subnetworks is adopted, to generate the semantic segmentation based representation, the 3D surface based representation, and the key point based rep-resentation of the human body, respectively. (ii) Based on the segmentation results, an appearance encoding branch is used to obtain the appearance information of the human body parts. (iii) The outputs of these two branches are represented by user-editable condition maps, which are then processed by a generator to predict the synthesized image. In this way, our BodyGAN can achieve the fine-grained dis-entanglement of pose, body shape, and appearance, and consequently enable the explicit and effective control of syn-thesis with diverse conditions. Extensive experiments on multiple datasets and a comprehensive user study show that our BodyGAN achieves the state-of-the-art performance.

Conventional visual relationship detection models only use the numeric ids of relation labels for training, but ignore the semantic correlation between the labels, which leads to severe training biases and harms the generalization ability of representations.In this paper, we introduce compact language information of relation labels for regularizing the representation learning of visual relations.Specifically, we propose a simple yet effective visual Relationship prediction framework that transfers natural language knowledge learned from Contrastive Language-Image Pre-training (CLIP) models to enhance the relationship prediction, termed as RelCLIP.Benefiting from the powerful visual-semantic alignment ability of CLIP at image level, we introduce a novel Relational Contrastive Learning (RCL) approach that explores relation-level visual-semantic alignment via learning to match cross-modal relational embeddings.By collaboratively learning the semantic coherence and discrepancy from relation triplets, the model can generate more discriminative and robust representations.Experimental results on the Visual Genome dataset show that RelCLIP achieves significant improvements over strong baselines under full (providing accurate labels) and distant supervision (providing noise labels), demonstrating its powerful generalization ability in learning relationship representations.

Geometry problem solving is a well-recognized testbed for evaluating the high-level multimodal reasoning capability of deep models.In most existing works, two main geometry problems: calculation and proving, are usually treated as two specific tasks, hindering a deep model to unify its reasoning capability on multiple math tasks.However, in essence, these two tasks have similar problem representations and overlapped math knowledge which can improve the understanding and reasoning ability of a deep model on both two tasks.Therefore, we construct a large-scale Unified Geometry problem benchmark, UniGeo, which contains 4,998 calculation problems and 9,543 proving problems.Each proving problem is annotated with a multi-step proof with reasons and mathematical expressions.The proof can be easily reformulated as a proving sequence that shares the same formats with the annotated program sequence for calculation problems.Naturally, we also present a unified multitask Geometric Transformer framework, Geoformer, to tackle calculation and proving problems simultaneously in the form of sequence generation, which finally shows the reasoning ability can be improved on both two tasks by unifying formulation.Furthermore, we propose a Mathematical Expression Pretraining (MEP) method that aims to predict the mathematical expressions in the problem solution, thus improving the Geoformer model.Experiments on the UniGeo demonstrate that our proposed Geoformer obtains state-of-the-art performance by outperforming task-specific model NGS with over 5.6% and 3.2% accuracies on calculation and proving problems, respectively.1

Providing Emotional Support (ES) to soothe people in emotional distress is an essential capability in social interactions.Most existing research on building ES conversation systems only considers single-turn interactions with users, which is over-simplified.In comparison, multi-turn ES conversation systems can provide ES more effectively, but face several new technical challenges, including: i) how to conduct support strategy planning that could lead to the best supporting effects; ii) how to dynamically model the user's state.In this paper, we propose a novel system named MultiESC to address these issues.For strategy planning, drawing inspiration from the A* search algorithm, we propose lookahead heuristics to estimate the future user feedback after using particular strategies, which helps to select strategies that can lead to the best long-term effects.For user state modeling, MultiESC focuses on capturing users' subtle emotional expressions and understanding their emotion causes.Extensive experiments show that MultiESC significantly outperforms competitive baselines in both strategy planning and dialogue generation.Our codes are available at https: //github.com/lwgkzl/MultiESC.

In this paper, we propose a comprehensive benchmark to investigate models' logical reasoning capabilities in complex real-life scenarios.Current explanation datasets often employ synthetic data with simple reasoning structures.Therefore, it cannot express more complex reasoning processes, such as the rebuttal to a reasoning step and the degree of certainty of the evidence.To this end, we propose a comprehensive logical reasoning explanation form.Based on the multi-hop chain of reasoning, the explanation form includes three main components:(1) The condition of rebuttal that the reasoning node can be challenged; (2) Logical formulae that uncover the internal texture of reasoning nodes; (3) Reasoning strength indicated by degrees of certainty.The fine-grained structure conforms to the real logical reasoning scenario, better fitting the human cognitive process but, simultaneously, is more challenging for the current models.We evaluate the current best models' performance on this new explanation form.The experimental results show that generating reasoning graphs remains a challenging task for current models, even with the help of giant pre-trained language models.

Cooperative multi-agent reinforcement learning (MARL) is making rapid progress for solving tasks in a grid world and real-world scenarios, in which agents are given different attributes and goals, resulting in different behavior through the whole multi-agent task. In this study, we quantify the agent’s behavior difference and build its relationship with the policy performance via {\bf Role Diversity}, a metric to measure the characteristics of MARL tasks. We define role diversity from three perspectives: action-based, trajectory-based, and contribution-based to fully measure a multi-agent task. Through theoretical analysis, we find that the error bound in MARL can be decomposed into three parts that have a strong relation to the role diversity. The decomposed factors can significantly impact policy optimization in three popular directions including parameter sharing, communication mechanism, and credit assignment. The main experimental platforms are based on {\bf Multiagent Particle Environment (MPE) }and {\bf The StarCraft Multi-Agent Challenge (SMAC)}. Extensive experiments clearly show that role diversity can serve as a robust measurement for the characteristics of a multi-agent cooperation task and help diagnose whether the policy fits the current multi-agent system for better policy performance.

With the rise of telemedicine, the task of developing Dialogue Systems for Medical Diagnosis (DSMD) has received much attention in recent years. Different from early researches that needed to rely on extra human resources and expertise to build the system, recent researches focused on how to build DSMD in a data-driven manner. However, the previous data-driven DSMD methods largely overlooked the system interpretability, which is critical for a medical application, and they also suffered from the data sparsity issue at the same time. In this paper, we explore how to bring interpretability to data-driven DSMD. Specifically, we propose a more interpretable decision process to implement the dialogue manager of DSMD by reasonably mimicking real doctors' inquiry logics, and we devise a model with highly transparent components to conduct the inference. Moreover, we collect a new DSMD dataset, which has a much larger scale, more diverse patterns, and is of higher quality than the existing ones. The experiments show that our method obtains 7.7%, 10.0%, 3.0% absolute improvement in diagnosis accuracy respectively on three datasets, demonstrating the effectiveness of its rational decision process and model design. Our codes and the GMD-12 dataset are available at https://github.com/lwgkzl/BR-Agent.

Cross-modal fashion image synthesis has emerged as one of the most promising directions in the generation domain due to the vast untapped potential of incorporating multiple modalities and the wide range of fashion image applications. To facilitate accurate generation, cross-modal synthesis methods typically rely on Contrastive Language-Image Pre-training (CLIP) to align textual and garment information. In this work, we argue that simply aligning texture and garment information is not sufficient to capture the semantics of the visual information and therefore propose MaskCLIP. MaskCLIP decomposes the garments into semantic parts, ensuring fine-grained and semantically accurate alignment between the visual and text information. Building on MaskCLIP, we propose ARMANI, a unified cross-modal fashion designer with part-level garment-text alignment. ARMANI discretizes an image into uniform tokens based on a learned cross-modal codebook in its first stage and uses a Transformer to model the distribution of image tokens for a real image given the tokens of the control signals in its second stage. Contrary to prior approaches that also rely on two-stage paradigms, ARMANI introduces textual tokens into the codebook, making it possible for the model to utilize fine-grain semantic information to generate more realistic images. Further, by introducing a cross-modal Transformer, ARMANI is versatile and can accomplish image synthesis from various control signals, such as pure text, sketch images, and partial images. Extensive experiments conducted on our newly collected cross-modal fashion dataset demonstrate that ARMANI generates photo-realistic images in diverse synthesis tasks and outperforms existing state-of-the-art cross-modal image synthesis approaches. Our code is available at https://github.com/Harvey594/ARMANI.

Guangyi Liu, Zichao Yang, Tianhua Tao, Xiaodan Liang, Junwei Bao, Zhen Li, Xiaodong He, Shuguang Cui, Zhiting Hu. Proceedings of the 2022 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies. 2022.

3D Garment modeling is a critical and challenging topic in the area of computer vision and graphics, with increasing attention focused on garment representation learning, garment reconstruction, and controllable garment manipulation, whereas existing methods were constrained to model garments under specific categories or with relatively simple topologies. In this paper, we propose a novel Neural Sewing Machine (NSM), a learning-based framework for structure-preserving 3D garment modeling, which is capable of learning representations for garments with diverse shapes and topologies and is successfully applied to 3D garment reconstruction and controllable manipulation. To model generic garments, we first obtain sewing pattern embedding via a unified sewing pattern encoding module, as the sewing pattern can accurately describe the intrinsic structure and the topology of the 3D garment. Then we use a 3D garment decoder to decode the sewing pattern embedding into a 3D garment using the UV-position maps with masks. To preserve the intrinsic structure of the predicted 3D garment, we introduce an inner-panel structure-preserving loss, an inter-panel structure-preserving loss, and a surface-normal loss in the learning process of our framework. We evaluate NSM on the public 3D garment dataset with sewing patterns with diverse garment shapes and categories. Extensive experiments demonstrate that the proposed NSM is capable of representing 3D garments under diverse garment shapes and topologies, realistically reconstructing 3D garments from 2D images with the preserved structure, and accurately manipulating the 3D garment categories, shapes, and topologies, outperforming the state-of-the-art methods by a clear margin.

Vision-Language Pre-training (VLP) models have shown remarkable performance on various downstream tasks. Their success heavily relies on the scale of pre-trained cross-modal datasets. However, the lack of large-scale datasets and benchmarks in Chinese hinders the development of Chinese VLP models and broader multilingual applications. In this work, we release a large-scale Chinese cross-modal dataset named Wukong, which contains 100 million Chinese image-text pairs collected from the web. Wukong aims to benchmark different multi-modal pre-training methods to facilitate the VLP research and community development. Furthermore, we release a group of models pre-trained with various image encoders (ViT-B/ViT-L/SwinT) and also apply advanced pre-training techniques into VLP such as locked-image text tuning, token-wise similarity in contrastive learning, and reduced-token interaction. Extensive experiments and a benchmarking of different downstream tasks including a new largest human-verified image-text test dataset are also provided. Experiments show that Wukong can serve as a promising Chinese pre-training dataset and benchmark for different cross-modal learning methods. For the zero-shot image classification task on 10 datasets, $Wukong_\text{ViT-L}$ achieves an average accuracy of 73.03%. For the image-text retrieval task, it achieves a mean recall of 71.6% on AIC-ICC which is 12.9% higher than WenLan 2.0. Also, our Wukong models are benchmarked on downstream tasks with other variants on multiple datasets, e.g., Flickr8K-CN, Flickr-30K-CN, COCO-CN, et al. More information can be referred to https://wukong-dataset.github.io/wukong-dataset/.

In this paper, we target image-based person-to-person virtual try-on in the presence of diverse poses and large viewpoint variations. Existing methods are restricted in this setting as they estimate garment warping flows mainly based on 2D poses and appearance, which omits the geometric prior of the 3D human body shape.Moreover, current garment warping methods are confined to localized regions, which makes them ineffective in capturing long-range dependencies and results in inferior flows with artifacts.To tackle these issues, we present 3D-aware global correspondences, which are reliable flows that jointly encode global semantic correlations, local deformations, and geometric priors of 3D human bodies. Particularly, given an image pair depicting the source and target person, (a) we first obtain their pose-aware and high-level representations via two encoders, and introduce a coarse-to-fine decoder with multiple refinement modules to predict the pixel-wise global correspondence. (b) 3D parametric human models inferred from images are incorporated as priors to regularize the correspondence refinement process so that our flows can be 3D-aware and better handle variations of pose and viewpoint. (c) Finally, an adversarial generator takes the garment warped by the 3D-aware flow, and the image of the target person as inputs, to synthesize the photo-realistic try-on result. Extensive experiments on public benchmarks and our selected HardPose test set demonstrate the superiority of our method against state-of-the-art try-on approaches.

Aiming towards a holistic understanding of multiple downstream tasks simultaneously, there is a need for extracting features with better transferability. Though many latest self-supervised pre-training methods have achieved impressive performance on various vision tasks under the prevailing pretrain-finetune paradigm, their generalization capacity to multi-task learning scenarios is yet to be explored. In this paper, we extensively investigate the transfer performance of various types of self-supervised methods, e.g., MoCo and SimCLR, on three downstream tasks, including semantic segmentation, drivable area segmentation, and traffic object detection, on the large-scale driving dataset BDD100K. We surprisingly find that their performances are sub-optimal or even lag far behind the single-task baseline, which may be due to the distinctions of training objectives and architectural design lied in the pretrain-finetune paradigm. To overcome this dilemma as well as avoid redesigning the resource-intensive pre-training stage, we propose a simple yet effective pretrain-adapt-finetune paradigm for general multi-task training, where the off-the-shelf pretrained models can be effectively adapted without increasing the training overhead. During the adapt stage, we utilize learnable multi-scale adapters to dynamically adjust the pretrained model weights supervised by multi-task objectives while leaving the pretrained knowledge untouched. Furthermore, we regard the vision-language pre-training model CLIP as a strong complement to the pretrain-adapt-finetune paradigm and propose a novel adapter named LV-Adapter, which incorporates language priors in the multi-task model via task-specific prompting and alignment between visual and textual features. Our experiments demonstrate that the adapt stage significantly improves the overall performance of those off-the-shelf pretrained models and the contextual features generated by LV-Adapter are of general benefits for downstream tasks.

Open-world object detection, as a more general and challenging goal, aims to recognize and localize objects described by arbitrary category names. The recent work GLIP formulates this problem as a grounding problem by concatenating all category names of detection datasets into sentences, which leads to inefficient interaction between category names. This paper presents DetCLIP, a paralleled visual-concept pre-training method for open-world detection by resorting to knowledge enrichment from a designed concept dictionary. To achieve better learning efficiency, we propose a novel paralleled concept formulation that extracts concepts separately to better utilize heterogeneous datasets (i.e., detection, grounding, and image-text pairs) for training. We further design a concept dictionary (with descriptions) from various online sources and detection datasets to provide prior knowledge for each concept. By enriching the concepts with their descriptions,we explicitly build the relationships among various concepts to facilitate the open-domain learning. The proposed concept dictionary is further used to provide sufficient negative concepts for the construction of the word-region alignment loss, and to complete labels for objects with missing descriptions in captions of image-text pair data. The proposed framework demonstrates strong zero-shot detection performances, e.g., on the LVIS dataset, our DetCLIP-T outperforms GLIP-T by 9.9% mAP and obtains a 13.5% improvement on rare categories compared to the fully-supervised model with the same backbone as ours.

Referring image segmentation aims at localizing all pixels of the visual objects described by a natural language sentence. Previous works learn to straightforwardly align the sentence embedding and pixel-level embedding for highlighting the referred objects, but ignore the semantic consistency of pixels within the same object, leading to incomplete masks and localization errors in predictions. To tackle this problem, we propose CoupAlign, a simple yet effective multi-level visual-semantic alignment method, to couple sentence-mask alignment with word-pixel alignment to enforce object mask constraint for achieving more accurate localization and segmentation. Specifically, the Word-Pixel Alignment (WPA) module performs early fusion of linguistic and pixel-level features in intermediate layers of the vision and language encoders. Based on the word-pixel aligned embedding, a set of mask proposals are generated to hypothesize possible objects. Then in the Sentence-Mask Alignment (SMA) module, the masks are weighted by the sentence embedding to localize the referred object, and finally projected back to aggregate the pixels for the target. To further enhance the learning of the two alignment modules, an auxiliary loss is designed to contrast the foreground and background pixels. By hierarchically aligning pixels and masks with linguistic features, our CoupAlign captures the pixel coherence at both visual and semantic levels, thus generating more accurate predictions. Extensive experiments on popular datasets (e.g., RefCOCO and G-Ref) show that our method achieves consistent improvements over state-of-the-art methods, e.g., about 2% oIoU increase on the validation and testing set of RefCOCO. Especially, CoupAlign has remarkable ability in distinguishing the target from multiple objects of the same class. Code will be available at https://gitee.com/mindspore/models/tree/master/research/cv/CoupAlign.

Prior highly-tuned image parsing models are usually studied in a certain domain with a specific set of semantic labels and can hardly be adapted into other scenarios (e.g.sharing discrepant label granularity) without extensive re-training. Learning a single universal parsing model by unifying label annotations from different domains or at various levels of granularity is a crucial but rarely addressed topic. This poses many fundamental learning challenges, e.g.discovering underlying semantic structures among different label granularity or mining label correlation across relevant tasks. To address these challenges, we propose a graph reasoning and transfer learning framework, named "Graphonomy," which incorporates human knowledge and label taxonomy into the intermediate graph representation learning beyond local convolutions. In particular, Graphonomy learns the global and structured semantic coherency in multiple domains via semantic-aware graph reasoning and transfer, enforcing the mutual benefits of the parsing across domains (e.g.different datasets or co-related tasks). The Graphonomy includes two iterated modules: Intra-Graph Reasoning and Inter-Graph Transfer modules. The former extracts the semantic graph in each domain to improve the feature representation learning by propagating information with the graph; the latter exploits the dependencies among the graphs from different domains for bidirectional knowledge transfer. We apply Graphonomy to two relevant but different image understanding research topics: human parsing and panoptic segmentation, and show Graphonomy can handle both of them well via a standard pipeline against current state-of-the-art approaches. Moreover, some extra benefit of our framework is demonstrated, e.g., generating the human parsing at various levels of granularity by unifying annotations across different datasets.

Language instruction plays an essential role in the natural language grounded navigation tasks. However, navigators trained with limited human-annotated instructions may have difficulties in accurately capturing key information from the complicated instruction at different timesteps, leading to poor navigation performance. In this paper, we exploit to train a more robust navigator which is capable of dynamically extracting crucial factors from the long instruction, by using an adversarial attacking paradigm. Specifically, we propose a Dynamic Reinforced Instruction Attacker (DR-Attacker), which learns to mislead the navigator to move to the wrong target by destroying the most instructive information in instructions at different timesteps. By formulating the perturbation generation as a Markov Decision Process, DR-Attacker is optimized by the reinforcement learning algorithm to generate perturbed instructions sequentially during the navigation, according to a learnable attack score. Then, the perturbed instructions, which serve as hard samples, are used for improving the robustness of the navigator with an effective adversarial training strategy and an auxiliary self-supervised reasoning task. Experimental results on both Vision-and-Language Navigation (VLN) and Navigation from Dialog History (NDH) tasks show the superiority of our proposed method over state-of-the-art methods. Moreover, the visualization analysis shows the effectiveness of the proposed DR-Attacker, which can successfully attack crucial information in the instructions at different timesteps. Code is available at https://github.com/expectorlin/DR-Attacker.

Beyond generating long and topic-coherent paragraphs in traditional captioning tasks, the medical image report composition task poses more task-oriented challenges by requiring both the highly accurate medical term diagnosis and multiple heterogeneous forms of information, including impression and findings. Current methods often generate the most common sentences due to dataset bias for the individual case, regardless of whether the sentences properly capture key entities and relationships. Such limitations severely hinder their applicability and generalization capability in medical report composition, where the most critical sentences lie in the descriptions of abnormal diseases that are relatively rare. Moreover, some medical terms appearing in one report are often entangled with each other and co-occurred, for example, symptoms associated with a specific disease. To enforce the semantic consistency of medical terms to be incorporated into the final reports and encourage the sentence generation for rare abnormal descriptions, we propose a novel framework that unifies template retrieval and sentence generation to handle both common and rare abnormality while ensuring the semantic coherency among the detected medical terms. Specifically, our approach exploits hybrid-knowledge co-reasoning: 1) explicit relationships among all abnormal medical terms to induce the visual attention learning and topic representation encoding for better topic-oriented symptoms descriptions and 2) adaptive generation mode that changes between the template retrieval and sentence generation according to a contextual topic encoder. The experimental results on two medical report benchmarks demonstrate the superiority of the proposed framework in terms of both human and metrics evaluation.

Though beneficial for encouraging the visual question answering (VQA) models to discover the underlying knowledge by exploiting the input-output correlation beyond image and text contexts, the existing knowledge VQA data sets are mostly annotated in a crowdsource way, e.g., collecting questions and external reasons from different users via the Internet. In addition to the challenge of knowledge reasoning, how to deal with the annotator bias also remains unsolved, which often leads to superficial overfitted correlations between questions and answers. To address this issue, we propose a novel data set named knowledge-routed visual question reasoning for VQA model evaluation. Considering that a desirable VQA model should correctly perceive the image context, understand the question, and incorporate its learned knowledge, our proposed data set aims to cut off the shortcut learning exploited by the current deep embedding models and push the research boundary of the knowledge-based visual question reasoning. Specifically, we generate the question-answer pair based on both the visual genome scene graph and an external knowledge base with controlled programs to disentangle the knowledge from other biases. The programs can select one or two triplets from the scene graph or knowledge base to push multistep reasoning, avoid answer ambiguity, and balance the answer distribution. In contrast to the existing VQA data sets, we further imply the following two major constraints on the programs to incorporate knowledge reasoning. First, multiple knowledge triplets can be related to the question, but only one knowledge relates to the image object. This can enforce the VQA model to correctly perceive the image instead of guessing the knowledge based on the given question solely. Second, all questions are based on different knowledge, but the candidate answers are the same for both the training and test sets. We make the testing knowledge unused during training to evaluate whether a model can understand question words and handle unseen combinations. Extensive experiments with various baselines and state-of-the-art VQA models are conducted to demonstrate that there still exists a big gap between the model with and without groundtruth supporting triplets when given the embedded knowledge base. This reveals the weakness of the current deep embedding models on the knowledge reasoning problem.

Visual commonsense knowledge has received growing attention in the reasoning of long-tailed visual relationships biased in terms of object and relation labels. Most current methods typically collect and utilize external knowledge for visual relationships by following the fixed reasoning path of {subject, object → predicate} to facilitate the recognition of infrequent relationships. However, the knowledge incorporation for such fixed multidependent path suffers from the data set biased and exponentially grown combinations of object and relation labels and ignores the semantic gap between commonsense knowledge and real scenes. To alleviate this, we propose configurable graph reasoning (CGR) to decompose the reasoning path of visual relationships and the incorporation of external knowledge, achieving configurable knowledge selection and personalized graph reasoning for each relation type in each image. Given a commonsense knowledge graph, CGR learns to match and retrieve knowledge for different subpaths and selectively compose the knowledge routed path. CGR adaptively configures the reasoning path based on the knowledge graph, bridges the semantic gap between the commonsense knowledge, and the real-world scenes and achieves better knowledge generalization. Extensive experiments show that CGR consistently outperforms previous state-of-the-art methods on several popular benchmarks and works well with different knowledge graphs. Detailed analyses demonstrated that CGR learned explainable and compelling configurations of reasoning paths.

When answering a question, people often draw upon their rich world knowledge in addition to the particular context. While recent works retrieve supporting facts/evidence from commonsense knowledge bases to supply additional information to each question, there is still ample opportunity to advance it on the quality of the evidence. It is crucial since the quality of the evidence is the key to answering common- sense questions, and even determines the upper bound on the QA systems’ performance. In this paper, we propose a recursive erasure memory network (REM-Net) to cope with the quality improvement of evidence. To address this, REM-Net is equipped with a module to refine the evidence by recursively erasing the low-quality evidence that does not explain the question answering. Besides, instead of retrieving evidence from existing knowledge bases, REM-Net leverages a pre-trained generative model to generate candidate evidence customized for the question. We conduct experiments on two commonsense question answering datasets, WIQA and CosmosQA. The results demonstrate the performance of REM- Net and show that the refined evidence is explainable.

Human doctors with well-structured medical knowledge can diagnose a disease merely via a few conversations with patients about symptoms. In contrast, existing knowledge-grounded dialogue systems often require a large number of dialogue instances to learn as they fail to capture the correlations between different diseases and neglect the diagnostic experience shared among them. To address this issue, we propose a more natural and practical paradigm, i.e., low-resource medical dialogue generation, which can transfer the diagnostic experience from source diseases to target ones with a handful of data for adaptation. It is capitalized on a commonsense knowledge graph to characterize the prior disease-symptom relations. Besides, we develop a Graph-Evolving Meta-Learning (GEML) framework that learns to evolve the commonsense graph for reasoning disease-symptom correlations in a new disease, which effectively alleviates the needs of a large number of dialogues. More importantly, by dynamically evolving disease-symptom graphs, GEML also well addresses the real-world challenges that the disease-symptom correlations of each disease may vary or evolve along with more diagnostic cases. Extensive experiment results on the CMDD dataset and our newly-collected Chunyu dataset testify the superiority of our approach over state-of-the-art approaches. Besides, our GEML can generate an enriched dialogue-sensitive knowledge graph in an online manner, which could benefit other tasks grounded on knowledge graph.

Low-resource automatic speech recognition (ASR) is challenging, as the low-resource target language data cannot well train an ASR model. To solve this issue, meta-learning formulates ASR for each source language into many small ASR tasks and meta-learns a model initialization on all tasks from different source languages to access fast adaptation on unseen target languages. However, for different source languages, the quantity and difficulty vary greatly because of their different data scales and diverse phonological systems, which leads to task-quantity and task-difficulty imbalance issues and thus a failure of multilingual meta-learning ASR (MML-ASR). In this work, we solve this problem by developing a novel adversarial meta sampling (AMS) approach to improve MML-ASR. When sampling tasks in MML-ASR, AMS adaptively determines the task sampling probability for each source language. Specifically, for each source language, if the query loss is large, it means that its tasks are not well sampled to train ASR model in terms of its quantity and difficulty and thus should be sampled more frequently for extra learning. Inspired by this fact, we feed the historical task query loss of all source language domain into a network to learn a task sampling policy for adversarially increasing the current query loss of MML-ASR. Thus, the learnt task sampling policy can master the learning situation of each language and thus predicts good task sampling probability for each language for more effective learning. Finally, experiment results on two multilingual datasets show significant performance improvement when applying our AMS on MML-ASR, and also demonstrate the applicability of AMS to other low-resource speech tasks and transfer learning ASR approaches.

Panoptic segmentation that unifies instance segmentation and semantic segmentation has recently attracted increasing attention. While most existing methods focus on designing novel architectures, we steer toward a different perspective: performing automated multi-loss adaptation (named Ada-Segment) on the fly to flexibly adjust multiple training losses over the course of training using a controller trained to capture the learning dynamics. This offers a few advantages: it bypasses manual tuning of the sensitive loss combination, a decisive factor for panoptic segmentation; allows to explicitly model the learning dynamics, and reconcile the learning of multiple objectives (up to ten in our experiments); with an end-to-end architecture, it generalizes to different datasets without the need of re-tuning hyperparameters or re-adjusting the training process laboriously. Our Ada-Segment brings 2.7% panoptic quality (PQ) improvement on COCO val split from the vanilla baseline, achieving the state-of-the-art 48.5% PQ on COCO test-dev split and 32.9% PQ on ADE20K dataset. The extensive ablation studies reveal the ever-changing dynamics throughout the training process, necessitating the incorporation of an automated and adaptive learning strategy as presented in this paper.

Pan Lu, Ran Gong, Shibiao Jiang, Liang Qiu, Siyuan Huang, Xiaodan Liang, Song-Chun Zhu. Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers). 2021.

Jinghui Qin, Xiaodan Liang, Yining Hong, Jianheng Tang, Liang Lin. Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers). 2021.

Zheng Ye, Liucun Lu, Lishan Huang, Liang Lin, Xiaodan Liang. Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers). 2021.

Recent breakthroughs of Neural Architecture Search (NAS) extend the field’s research scope towards a broader range of vision tasks and more diversified search spaces. While existing NAS methods mostly design architectures on a single task, algorithms that look beyond single-task search are surging to pursue a more efficient and universal solution across various tasks. Many of them leverage transfer learning and seek to preserve, reuse, and refine network design knowledge to achieve higher efficiency in future tasks. However, the enormous computational cost and experiment complexity of cross-task NAS are imposing barriers for valuable research in this direction. Existing NAS benchmarks all focus on one type of vision task, i.e., classification. In this work, we propose TransNAS-Bench-101, a benchmark dataset containing network performance across seven tasks, covering classification, regression, pixel-level prediction, and self-supervised tasks. This diversity provides opportunities to transfer NAS methods among tasks and allows for more complex transfer schemes to evolve. We explore two fundamentally different types of search space: cell-level search space and macro-level search space. With 7,352 backbones evaluated on seven tasks, 51,464 trained models with detailed training information are provided. With TransNAS-Bench-101, we hope to encourage the advent of exceptional NAS algorithms that raise cross-task search efficiency and generalizability to the next level. Our dataset and code will be available at Mindspore1and VEGA2.

Current dynamic networks and dynamic pruning methods have shown their promising capability in reducing theoretical computation complexity. However, dynamic sparse patterns on convolutional filters fail to achieve actual acceleration in real-world implementation, due to the extra burden of indexing, weight-copying, or zero-masking. Here, we explore a dynamic network slimming regime, named Dynamic Slimmable Network (DS-Net), which aims to achieve good hardware-efficiency via dynamically adjusting filter numbers of networks at test time with respect to different inputs, while keeping filters stored statically and contiguously in hardware to prevent the extra burden. Our DS-Net is empowered with the ability of dynamic inference by the proposed double-headed dynamic gate that comprises an attention head and a slimming head to predictively adjust network width with negligible extra computation cost. To ensure generality of each candidate architecture and the fairness of gate, we propose a disentangled two-stage training scheme inspired by one-shot NAS. In the first stage, a novel training technique for weight-sharing networks named In-place Ensemble Bootstrapping is proposed to improve the supernet training efficacy. In the second stage, Sandwich Gate Sparsification is proposed to assist the gate training by identifying easy and hard samples in an online way. Extensive experiments demonstrate our DS-Net consistently outperforms its static counterparts as well as state-of-the-art static and dynamic model compression methods by a large margin (up to 5.9%). Typically, DS-Net achieves 2-4× computation reduction and 1.62× real-world acceleration over ResNet-50 and MobileNet with minimal accuracy drops on ImageNet.1

The ability to navigate like a human towards a language-guided target from anywhere in a 3D embodied environment is one of the ‘holy grail’ goals of intelligent robots. Most visual navigation benchmarks, however, focus on navigating toward a target from a fixed starting point, guided by an elaborate set of instructions that depicts step-by-step. This approach deviates from real-world problems in which human-only describes what the object and its surrounding look like and asks the robot to start navigation from any-where. Accordingly, in this paper, we introduce a Scenario Oriented Object Navigation (SOON) task. In this task, an agent is required to navigate from an arbitrary position in a 3D embodied environment to localize a target following a scene description. To give a promising direction to solve this task, we propose a novel graph-based exploration (GBE) method, which models the navigation state as a graph and introduces a novel graph-based exploration approach to learn knowledge from the graph and stabilize training by learning sub-optimal trajectories. We also propose a new large-scale benchmark named From Anywhere to Object (FAO) dataset. To avoid target ambiguity, the descriptions in FAO provide rich semantic scene information includes: object attribute, object relationship, region description, and nearby region description. Our experiments reveal that the proposed GBE outperforms various state-of-the-arts on both FAO and R2R datasets. And the ablation studies on FAO validates the quality of the dataset.

Vision-language Navigation (VLN) tasks require an agent to navigate step-by-step while perceiving the visual observations and comprehending a natural language instruction. Large data bias, which is caused by the disparity ratio between the small data scale and large navigation space, makes the VLN task challenging. Previous works have proposed various data augmentation methods to reduce data bias. However, these works do not explicitly reduce the data bias across different house scenes. Therefore, the agent would overfit to the seen scenes and achieve poor navigation performance in the unseen scenes. To tackle this problem, we propose the Random Environmental Mixup (REM) method, which generates cross-connected house scenes as augmented data via mixuping environment. Specifically, we first select key viewpoints according to the room connection graph for each scene. Then, we cross-connect the key views of different scenes to construct augmented scenes. Finally, we generate augmented instruction-path pairs in the cross-connected scenes. The experimental results on benchmark datasets demonstrate that our augmentation data via REM help the agent reduce its performance gap between the seen and unseen environment and improve the overall performance, making our model the best existing approach on the standard VLN benchmark.

Vision-Dialog Navigation (VDN) requires an agent to ask questions and navigate following the human responses to find target objects. Conventional approaches are only allowed to ask questions at predefined locations, which are built upon expensive dialogue annotations, and inconvenience the real-word human-robot communication and cooperation. In this paper, we propose a Self-Motivated Communication Agent (SCoA) that learns whether and what to communicate with human adaptively to acquire instructive information for realizing dialogue annotation-free navigation and enhancing the transferability in real-world unseen environment. Specifically, we introduce a whether-to-ask (WeTA) policy, together with uncertainty of which action to choose, to indicate whether the agent should ask a question. Then, a what-to-ask (WaTA) policy is proposed, in which, along with the oracle’s answers, the agent learns to score question candidates so as to pick up the most informative one for navigation, and meanwhile mimic oracle’s answering. Thus, the agent can navigate in a self-Q&A manner even in real-world environment where the human assistance is often unavailable. Through joint optimization of communication and navigation in a unified imitation learning and reinforcement learning framework, SCoA asks a question if necessary and obtains a hint for guiding the agent to move towards the target with less communication cost. Experiments on seen and unseen environments demonstrate that SCoA shows not only superior performance over existing baselines without dialog annotations, but also competing results compared with rich dialog annotations based counterparts.

Generalization on out-of-distribution (OOD) test data is an essential but underexplored topic in visual question answering. Current state-of-the-art VQA models often exploit the biased correlation between data and labels, which results in a large performance drop when the test and training data have different distributions. Inspired by the fact that humans can recognize novel concepts by composing existed concepts and capsule network’s ability of representing part-whole hierarchies, we propose to use capsules to represent parts and introduce "Linguistically Routing" to merge parts with human-prior hierarchies. Specifically, we first fuse visual features with a single question word as atomic parts. Then we introduce the "Linguistically Routing" to reweight the capsule connections between two layers such that: 1) the lower layer capsules can transfer their outputs to the most compatible higher capsules, and 2) two capsules can be merged if their corresponding words are merged in the question parse tree. The routing process maximizes the above unary and binary potentials across multiple layers and finally carves a tree structure inside the capsule network. We evaluate our proposed routing method on the CLEVR compositional generation test, the VQA-CP2 dataset and the VQAv2 dataset. The experimental results show that our proposed method can improve current VQA models on OOD split without losing performance on the in-domain test data.

A myriad of recent breakthroughs in hand-crafted neural architectures for visual recognition have highlighted the urgent need to explore hybrid architectures consisting of diversified building blocks. Meanwhile, neural architecture search methods are surging with an expectation to reduce human efforts. However, whether NAS methods can efficiently and effectively handle diversified search spaces with disparate candidates (e.g. CNNs and transformers) is still an open question. In this work, we present Block-wisely Self-supervised Neural Architecture Search (BossNAS), an unsupervised NAS method that addresses the problem of in-accurate architecture rating caused by large weight-sharing space and biased supervision in previous methods. More specifically, we factorize the search space into blocks and utilize a novel self-supervised training scheme, named ensemble bootstrapping, to train each block separately before searching them as a whole towards the population center. Additionally, we present HyTra search space, a fabric-like hybrid CNN-transformer search space with searchable down-sampling positions. On this challenging search space, our searched model, BossNet-T, achieves up to 82.5% accuracy on ImageNet, surpassing EfficientNet by 2.4% with comparable compute time. Moreover, our method achieves superior architecture rating accuracy with 0.78 and 0.76 Spearman correlation on the canonical MBConv search space with ImageNet and on NATS-Bench size search space with CIFAR-100, respectively, surpassing state-of-the-art NAS methods.1

Current 3D object detection paradigms highly rely on extensive annotation efforts, which makes them not practical in many real-world industrial applications. Inspired by that a human driver can keep accumulating experiences from self-exploring the roads without any tutor’s guidance, we first step forwards to explore a simple yet effective self-supervised learning framework tailored for LiDAR-based 3D object detection. Although the self-supervised pipeline has achieved great success in 2D domain, the characteristic challenges (e.g., complex geometry structure and various 3D object views) encountered in the 3D domain hinder the direct adoption of existing techniques that often contrast the 2D augmented data or cluster single-view features. Here we present a novel self-supervised 3D Object detection framework that seamlessly integrates the geometry-aware contrast and clustering harmonization to lift the unsupervised 3D representation learning, named GCC-3D. First, GCC-3D introduces a Geometric-Aware Contrastive objective to learn spatial-sensitive local structure representation. This objective enforces the spatially close voxels to have high feature similarity. Second, a Pseudo-Instance Clustering harmonization mechanism is proposed to encourage that different views of pseudo-instances should have consistent similarities to clustering prototype centers. This module endows our model semantic discriminative capacity. Extensive experiments demonstrate our GCC-3D achieves significant performance improvement on data-efficient 3D object detection benchmarks (nuScenes and Waymo). Moreover, our GCC-3D framework can achieve state-of-the art performances on all popular 3D object detection benchmarks.

Knowledge Distillation has shown very promising ability in transferring learned representation from the larger model (teacher) to the smaller one (student). Despite many efforts, prior methods ignore the important role of retaining inter-channel correlation of features, leading to the lack of capturing intrinsic distribution of the feature space and sufficient diversity properties of features in the teacher network. To solve the issue, we propose the novel Inter-Channel Correlation for Knowledge Distillation (ICKD), with which the diversity and homology of the feature space of the student network can align with that of the teacher network. The correlation between these two channels is interpreted as diversity if they are irrelevant to each other, otherwise homology. Then the student is required to mimic the correlation within its own embedding space. In addition, we introduce the grid-level inter-channel correlation, making it capable of dense prediction tasks. Extensive experiments on two vision tasks, including ImageNet classification and Pascal VOC segmentation, demonstrate the superiority of our ICKD, which consistently outperforms many existing methods, advancing the state-of-the-art in the fields of Knowledge Distillation. To our knowledge, we are the first method based on knowledge distillation boosts ResNet18 beyond 72% Top-1 accuracy on ImageNet classification. Code is available at: https://github.com/ADLab-AutoDrive/ICKD.

We present a flexible and high-performance framework, named Pyramid R-CNN, for two-stage 3D object detection from point clouds. Current approaches generally rely on the points or voxels of interest for RoI feature extraction on the second stage, but cannot effectively handle the sparsity and non-uniform distribution of those points, and this may result in failures in detecting objects that are far away. To resolve the problems, we propose a novel second-stage module, named pyramid RoI head, to adaptively learn the features from the sparse points of interest. The pyramid RoI head consists of three key components. Firstly, we propose the RoI-grid Pyramid, which mitigates the sparsity problem by extensively collecting points of interest for each RoI in a pyramid manner. Secondly, we propose RoI-grid Attention, a new operation that can encode richer information from sparse points by incorporating conventional attention-based and graph-based point operators into a unified formulation. Thirdly, we propose the Density-Aware Radius Prediction (DARP) module, which can adapt to different point density levels by dynamically adjusting the focusing range of RoIs. Combining the three components, our pyramid RoI head is robust to the sparse and imbalanced circumstances, and can be applied upon various 3D backbones to consistently boost the detection performance. Extensive experiments show that Pyramid R-CNN outperforms the state-of-the-art 3D detection models by a large margin on both the KITTI dataset and the Waymo Open dataset.

We present Voxel Transformer (VoTr), a novel and effective voxel-based Transformer backbone for 3D object detection from point clouds. Conventional 3D convolutional backbones in voxel-based 3D detectors cannot efficiently capture large context information, which is crucial for object recognition and localization, owing to the limited receptive fields. In this paper, we resolve the problem by introducing a Transformer-based architecture that enables long-range relationships between voxels by self-attention. Given the fact that non-empty voxels are naturally sparse but numerous, directly applying standard Transformer on voxels is non-trivial. To this end, we propose the sparse voxel module and the submanifold voxel module, which can operate on the empty and non-empty voxel positions effectively. To further enlarge the attention range while maintaining comparable computational overhead to the convolutional counterparts, we propose two attention mechanisms for multi-head attention in those two modules: Local Attention and Dilated Attention, and we further propose Fast Voxel Query to accelerate the querying process in multi-head attention. VoTr contains a series of sparse and submanifold voxel modules, and can be applied in most voxel-based detectors. Our proposed VoTr shows consistent improvement over the convolutional baselines while maintaining computational efficiency on the KITTI dataset and the Waymo Open dataset.

Recently proposed neural architecture search (NAS) methods co-train billions of architectures in a supernet and estimate their potential accuracy using the network weights detached from the supernet. However, the ranking correlation between the architectures' predicted accuracy and their actual capability is incorrect, which causes the existing NAS methods' dilemma. We attribute this ranking correlation problem to the supernet training consistency shift, including feature shift and parameter shift. Feature shift is identified as dynamic input distributions of a hidden layer due to random path sampling. The input distribution dynamic affects the loss descent and finally affects architecture ranking. Parameter shift is identified as contradictory parameter updates for a shared layer lay in different paths in different training steps. The rapidly-changing parameter could not preserve architecture ranking. We address these two shifts simultaneously using a nontrivial supernet-∏ model, called ∏-NAS. Specifically, we employ a supernet-∏ model that contains cross-path learning to reduce the feature consistency shift between different paths. Meanwhile, we adopt a novel nontrivial mean teacher containing negative samples to overcome parameter shift and model collision. Furthermore, our ∏-NAS runs in an unsupervised manner, which can search for more transferable architectures. Extensive experiments on ImageNet and a wide range of downstream tasks (e.g., COCO 2017, ADE20K, and Cityscapes) demonstrate the effectiveness and universality of our ∏-NAS compared to supervised NAS. See Codes1.

Fine-tuning from pre-trained ImageNet models has been a simple, effective, and popular approach for various computer vision tasks. The common practice of fine-tuning is to adopt a default hyperparameter setting with a fixed pre-trained model, while both of them are not optimized for specific tasks and time constraints. Moreover, in cloud computing or GPU clusters where the tasks arrive sequentially in a stream, faster online fine-tuning is a more desired and realistic strategy for saving money, energy consumption, and CO2 emission. In this paper, we propose a joint Neural Architecture Search and Online Adaption framework named NASOA towards a faster task-oriented fine-tuning upon the request of users. Specifically, NASOA first adopts an offline NAS to identify a group of training-efficient networks to form a pretrained model zoo. We propose a novel joint block and macro level search space to enable a flexible and effi-cient search. Then, by estimating fine-tuning performance via an adaptive model by accumulating experience from the past tasks, an online schedule generator is proposed to pick up the most suitable model and generate a personalized training regime with respect to each desired task in a one-shot fashion. The resulting model zoo1is more training efficient than SOTA models, e.g. 6x faster than RegNetY-16GF, and 1.7x faster than EfficientNetB3. Experiments on multiple datasets also show that NASOA achieves much better fine-tuning results, i.e. improving around 2.1% accuracy than the best performance in RegNet series under various constraints and tasks; 40x faster compared to the BOHB.

Recovering dense human poses from images plays a critical role in establishing an image-to-surface correspondence between RGB images and the 3D surface of the human body, serving the foundation of rich real-world applications, such as virtual humans, monocular-to-3d reconstruction. However, the popular DensePose-COCO dataset relies on a sophisticated manual annotation system, leading to severe limitations in acquiring the denser and more accurate annotated pose resources. In this work, we introduce a new 3D human-body model with a series of decoupled parameters that could freely control the generation of the body. Furthermore, we build a data generation system based on this decoupling 3D model, and construct an ultra dense synthetic benchmark UltraPose, containing around 1.3 billion corresponding points. Compared to the existing manually annotated DensePose-COCO dataset, the synthetic UltraPose has ultra dense image-to-surface correspondences without annotation cost and error. Our proposed UltraPose provides the largest benchmark and data resources for lifting the model capability in predicting more accurate dense poses. To promote future researches in this field, we also propose a transformer-based method to model the dense correspondence between 2D and 3D worlds. The proposed model trained on synthetic UltraPose can be applied to real-world scenarios, indicating the effectiveness of our benchmark and model.1

Nowadays, customer’s demands for E-commerce are more diversified, which introduces more complications to the product retrieval industry. Previous methods are either subject to single-modal input or perform supervised image-level product retrieval, thus fail to accommodate real-life scenarios where enormous weakly annotated multi-modal data are present. In this paper, we investigate a more realistic setting that aims to perform weakly-supervised multi-modal instance-level product retrieval among fine-grained product categories. To promote the study of this challenging task, we contribute Product1M, one of the largest multi-modal cosmetic datasets for real-world instance-level retrieval. Notably, Product1M contains over 1 million image-caption pairs and consists of two sample types, i.e., single-product and multi-product samples, which encompass a wide variety of cosmetics brands. In addition to the great diversity, Product1M enjoys several appealing characteristics including fine-grained categories, complex combinations, and fuzzy correspondence that well mimic the real-world scenes. Moreover, we propose a novel model named Cross-modal contrAstive Product Transformer for instance-level prodUct REtrieval (CAPTURE), that excels in capturing the potential synergy between multi-modal inputs via a hybrid-stream transformer in a self-supervised manner. CAPTURE generates discriminative instance features via masked multi-modal learning as well as cross-modal contrastive pretraining and it outperforms several SOTA cross-modal baselines. Extensive ablation studies well demonstrate the effectiveness and the generalization capacity of our model. Dataset and codes are available at https: //github.com/zhanxlin/Product1M.

Virtual 3D try-on can provide an intuitive and realistic view for online shopping and has a huge potential commercial value. However, existing 3D virtual try-on methods mainly rely on annotated 3D human shapes and garment templates, which hinders their applications in practical scenarios. 2D virtual try-on approaches provide a faster alternative to manipulate clothed humans, but lack the rich and realistic 3D representation. In this paper, we propose a novel Monocular-to-3D Virtual Try-On Network (M3D-VTON) that builds on the merits of both 2D and 3D approaches. By integrating 2D information efficiently and learning a mapping that lifts the 2D representation to 3D, we make the first attempt to reconstruct a 3D try-on mesh only taking the target clothing and a person image as inputs. The proposed M3D-VTON includes three modules: 1) The Monocular Prediction Module (MPM) that estimates an initial full-body depth map and accomplishes 2D clothes-person alignment through a novel two-stage warping procedure; 2) The Depth Refinement Module (DRM) that refines the initial body depth to produce more detailed pleat and face characteristics; 3) The Texture Fusion Module (TFM) that fuses the warped clothing with the non-target body part to refine the results. We also construct a high-quality synthesized Monocular-to-3D virtual try-on dataset, in which each person image is associated with a front and a back depth map. Extensive experiments demonstrate that the proposed M3D-VTON can manipulate and reconstruct the 3D human body wearing the given clothing with compelling details and is more efficient than other 3D approaches.1

Network architecture search techniques have recently revolutionized massive application problems due to its advanced performance. However, previous works rely on a strong assumption that the optimizers are manually designed and fixed for all architectures, which deviates from the intuition that each architecture should be empowered with its optimal optimization strategy. In this paper, we propose to unify Dynamic Optimizer Search (DOS) and Network Architecture Search (NAS), aiming to jointly search for the optimization scheduling and network architecture, named as DOS-NAS. The scheduling process of DOS is formulated as a Markov Decision Process, optimized via the REINFORCE algorithm. Conjunct with NAS, our DOS-NAS learns a discrete temporal scheduling to dynamically configure optimizers from a predefined set, leading to personalized optimizer scheduling for the searched network. Extensive experiments on CIFAR-10, CIFAR-100 and Tiny-ImageNet demonstrate the validity of our proposed DOS-NAS in performance improvement and speeding up the convergence significantly. DOS-NAS provides a new paradigm for AutoML systems through the joint search over optimization strategy and network architecture.

Transformer-based models are popularly used in natural language processing (NLP). Its core component, self-attention, has aroused widespread interest. To understand the self-attention mechanism, a direct method is to visualize the attention map of a pre-trained model. Based on the patterns observed, a series of efficient Transformers with different sparse attention masks have been proposed. From a theoretical perspective, universal approximability of Transformer-based models is also recently proved. However, the above understanding and analysis of self-attention is based on a pre-trained model. To rethink the importance analysis in self-attention, we study the significance of different positions in attention matrix during pre-training. A surprising result is that diagonal elements in the attention map are the least important compared with other attention positions. We provide a proof showing that these diagonal elements can indeed be removed without deteriorating model performance. Furthermore, we propose a Differentiable Attention Mask (DAM) algorithm, which further guides the design of the SparseBERT. Extensive experiments verify our interesting findings and illustrate the effect of the proposed algorithm.

Although deep reinforcement learning (RL) has been successfully applied to a variety of robotic control tasks, it’s still challenging to apply it to real-world tasks, due to the poor sample efficiency. Attempting to overcome this shortcoming, several works focus on reusing the collected trajectory data during the training by decomposing them into a set of policy-irrelevant discrete transitions. However, their improvements are somewhat marginal since i) the amount of the transitions is usually small, and ii) the value assignment only happens in the joint states. To address these issues, this paper introduces a concise yet powerful method to construct Continuous Transition, which exploits the trajectory information by exploiting the potential transitions along the trajectory. Specifically, we propose to synthesize new transitions for training by linearly interpolating the consecutive transitions. To keep the constructed transitions authentic, we also develop a discriminator to guide the construction process automatically. Extensive experiments demonstrate that our proposed method achieves a significant improvement in sample efficiency on various complex continuous robotic control problems in MuJoCo and outperforms the advanced model-based / model-free RL methods. The source code is available1.

Despite recent progress on image-based virtual try-on, current methods are constraint by shared warping networks and thus fail to synthesize natural try-on results when faced with clothing categories that require different warping operations. In this paper, we address this problem by finding clothing category-specific warping networks for the virtual try-on task via Neural Architecture Search (NAS). We introduce a NAS-Warping Module and elaborately design a bilevel hierarchical search space to identify the optimal network-level and operation-level flow estimation architecture. Given the network-level search space, containing different numbers of warping blocks, and the operation-level search space with different convolution operations, we jointly learn a combination of repeatable warping cells and convolution operations specifically for the clothing-person alignment. Moreover, a NAS-Fusion Module is proposed to synthesize more natural final try-on results, which is realized by leveraging particular skip connections to produce better-fused features that are required for seamlessly fusing the warped clothing and the unchanged person part. We adopt an efficient and stable one-shot searching strategy to search the above two modules. Extensive experiments demonstrate that our WAS-VTON significantly outperforms the previous fixed-architecture try-on methods with more natural warping results and virtual try-on results.

Yinya Huang, Meng Fang, Yu Cao, Liwei Wang, Xiaodan Liang. Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies. 2021.

Image-based virtual try-on is one of the most promising applications of human-centric image generation due to its tremendous real-world potential. Yet, as most try-on approaches fit in-shop garments onto a target person, they require the laborious and restrictive construction of a paired training dataset, severely limiting their scalability. While a few recent works attempt to transfer garments directly from one person to another, alleviating the need to collect paired datasets, their performance is impacted by the lack of paired (supervised) information. In particular, disentangling style and spatial information of the garment becomes a challenge, which existing methods either address by requiring auxiliary data or extensive online optimization procedures, thereby still inhibiting their scalability. To achieve a scalable virtual try-on system that can transfer arbitrary garments between a source and a target person in an unsupervised manner, we thus propose a texture-preserving end-to-end network, the PAtch-routed SpaTially-Adaptive GAN (PASTA-GAN), that facilitates real-world unpaired virtual try-on. Specifically, to disentangle the style and spatial information of each garment, PASTA-GAN consists of an innovative patch-routed disentanglement module for successfully retaining garment texture and shape characteristics. Guided by the source person's keypoints, the patch-routed disentanglement module first decouples garments into normalized patches, thus eliminating the inherent spatial information of the garment, and then reconstructs the normalized patches to the warped garment complying with the target person pose. Given the warped garment, PASTA-GAN further introduces novel spatially-adaptive residual blocks that guide the generator to synthesize more realistic garment details. Extensive comparisons with paired and unpaired approaches demonstrate the superiority of PASTA-GAN, highlighting its ability to generate high-quality try-on images when faced with a large variety of garments(e.g. vests, shirts, pants), taking a crucial step towards real-world scalable try-on.

Collaborative reasoning for understanding image-question pairs is a very critical but underexplored topic in interpretable visual question answering systems. Although very recent studies have attempted to use explicit compositional processes to assemble multiple subtasks embedded in questions, their models heavily rely on annotations or handcrafted rules to obtain valid reasoning processes, which leads to either heavy workloads or poor performance on compositional reasoning. In this paper, to better align image and language domains in diverse and unrestricted cases, we propose a novel neural network model that performs global reasoning on a dependency tree parsed from the question; thus, our model is called a parse-tree-guided reasoning network (PTGRN). This network consists of three collaborative modules: i) an attention module that exploits the local visual evidence of each word parsed from the question, ii) a gated residual composition module that composes the previously mined evidence, and iii) a parse-tree-guided propagation module that passes the mined evidence along the parse tree. Thus, PTGRN is capable of building an interpretable visual question answering (VQA) system that gradually derives image cues following question-driven parse-tree reasoning. Experiments on relational datasets demonstrate the superiority of PTGRN over current state-of-the-art VQA methods, and the visualization results highlight the explainable capability of our reasoning system.

Transferring human motion from a source to a target person poses great potential in computer vision and graphics applications. A crucial step is to manipulate sequential future motion while retaining the appearance characteristic. Previous work has either relied on crafted 3D human models or trained a separate model specifically for each target person, which is not scalable in practice. This work studies a more general setting, in which we aim to learn a single model to parsimoniously transfer motion from a source video to any target person given only one image of the person, named as Collaborative Parsing-Flow Network (CPF-Net). The paucity of information regarding the target person makes the task particularly challenging to faithfully preserve the appearance in varying designated poses. To address this issue, CPF-Net integrates the structured human parsing and appearance flow to guide the realistic foreground synthesis which is merged into the background by a spatio-temporal fusion module. In particular, CPF-Net decouples the problem into stages of human parsing sequence generation, foreground sequence generation and final video generation. The human parsing generation stage captures both the pose and the body structure of the target. The appearance flow is beneficial to keep details in synthesized frames. The integration of human parsing and appearance flow effectively guides the generation of video frames with realistic appearance. Finally, the dedicated designed fusion network ensure the temporal coherence. We further collect a large set of human dancing videos to push forward this research field. Both quantitative and qualitative results show our method substantially improves over previous approaches and is able to generate appealing and photo-realistic target videos given any input person image. All source code and dataset will be released at https://github.com/xiezhy6/CPF-Net.

Non-parallel text style transfer has attracted increasing research interests in recent years. Despite successes in transferring the style based on the encoder-decoder framework, current approaches still lack the ability to preserve the content and even logic of original sentences, mainly due to the large unconstrained model space or too simplified assumptions on latent embedding space. Since language itself is an intelligent product of humans with certain grammars and has a limited rule-based model space by its nature, relieving this problem requires reconciling the model capacity of deep neural networks with the intrinsic model constraints from human linguistic rules. To this end, we propose a method called Graph Transformer–based Auto-Encoder, which models a sentence as a linguistic graph and performs feature extraction and style transfer at the graph level, to maximally retain the content and the linguistic structure of original sentences. Quantitative experiment results on three non-parallel text style transfer tasks show that our model outperforms state-of-the-art methods in content preservation, while achieving comparable performance on transfer accuracy and sentence naturalness.

Medical imaging technologies, including computed tomography (CT) or chest X-Ray (CXR), are largely employed to facilitate the diagnosis of the COVID-19. Since manual report writing is usually too time-consuming, a more intelligent auxiliary medical system that could generate medical reports automatically and immediately is urgently needed. In this article, we propose to use the medical visual language BERT (Medical-VLBERT) model to identify the abnormality on the COVID-19 scans and generate the medical report automatically based on the detected lesion regions. To produce more accurate medical reports and minimize the visual-and-linguistic differences, this model adopts an alternate learning strategy with two procedures that are knowledge pretraining and transferring. To be more precise, the knowledge pretraining procedure is to memorize the knowledge from medical texts, while the transferring procedure is to utilize the acquired knowledge for professional medical sentences generations through observations of medical images. In practice, for automatic medical report generation on the COVID-19 cases, we constructed a dataset of 368 medical findings in Chinese and 1104 chest CT scans from The First Affiliated Hospital of Jinan University, Guangzhou, China, and The Fifth Affiliated Hospital of Sun Yat-sen University, Zhuhai, China. Besides, to alleviate the insufficiency of the COVID-19 training samples, our model was first trained on the large-scale Chinese CX-CHR dataset and then transferred to the COVID-19 CT dataset for further fine-tuning. The experimental results showed that Medical-VLBERT achieved state-of-the-art performances on terminology prediction and report generation with the Chinese COVID-19 CT dataset and the CX-CHR dataset. The Chinese COVID-19 CT dataset is available at https://covid19ct.github.io/.

Most advances in medical lesion detection network are limited to subtle modification on the conventional detection network designed for natural images. However, there exists a vast domain gap between medical images and natural images where the medical image detection often suffers from several domain-specific challenges, such as high lesion/background similarity, dominant tiny lesions, and severe class imbalance. Is a hand-crafted detection network tailored for natural image undoubtedly good enough over a discrepant medical lesion domain? Is there more powerful operations, filters, and sub-networks that better fit the medical lesion detection problem to be discovered? In this paper, we introduce a novel ElixirNet that includes three components: 1) TruncatedRPN balances positive and negative data for false positive reduction; 2) Auto-lesion Block is automatically customized for medical images to incorporates relation-aware operations among region proposals, and leads to more suitable and efficient classification and localization. 3) Relation transfer module incorporates the semantic relationship and transfers the relevant contextual information with an interpretable graph, thus alleviates the problem of lack of annotations for all types of lesions. Experiments on DeepLesion and Kits19 prove the effectiveness of ElixirNet, achieving improvement of both sensitivity and precision over FPN with fewer parameters.

Target-guided open-domain conversation aims to proactively and naturally guide a dialogue agent or human to achieve specific goals, topics or keywords during open-ended conversations. Existing methods mainly rely on single-turn data-driven learning and simple target-guided strategy without considering semantic or factual knowledge relations among candidate topics/keywords. This results in poor transition smoothness and low success rate. In this work, we adopt a structured approach that controls the intended content of system responses by introducing coarse-grained keywords, attains smooth conversation transition through turn-level supervised learning and knowledge relations between candidate keywords, and drives an conversation towards an specified target with discourse-level guiding strategy. Specially, we propose a novel dynamic knowledge routing network (DRKN) which considers semantic knowledge relations among candidate keywords for accurate next topic prediction of next discourse. With the help of more accurate keyword prediction, our keyword-augmented response retrieval module can achieve better retrieval performance and more meaningful conversations. Besides, we also propose a novel dual discourse-level target-guided strategy to guide conversations to reach their goals smoothly with higher success rate. Furthermore, to push the research boundary of target-guided open-domain conversation to match real-world scenarios better, we introduce a new large-scale Chinese target-guided open-domain conversation dataset (more than 900K conversations) crawled from Sina Weibo. Quantitative and human evaluations show our method can produce meaningful and effective target-guided conversations, significantly improving over other state-of-the-art methods by more than 20% in success rate and more than 0.6 in average smoothness score.

The dominant object detection approaches treat each dataset separately and fit towards a specific domain, which cannot adapt to other domains without extensive retraining. In this paper, we address the problem of designing a universal object detection model that exploits diverse category granularity from multiple domains and predict all kinds of categories in one system. Existing works treat this problem by integrating multiple detection branches upon one shared backbone network. However, this paradigm overlooks the crucial semantic correlations between multiple domains, such as categories hierarchy, visual similarity, and linguistic relationship. To address these drawbacks, we present a novel universal object detector called Universal-RCNN that incorporates graph transfer learning for propagating relevant semantic information across multiple datasets to reach semantic coherency. Specifically, we first generate a global semantic pool by integrating all high-level semantic representation of all the categories. Then an Intra-Domain Reasoning Module learns and propagates the sparse graph representation within one dataset guided by a spatial-aware GCN. Finally, an Inter-Domain Transfer Module is proposed to exploit diverse transfer dependencies across all domains and enhance the regional feature representation by attending and transferring semantic contexts globally. Extensive experiments demonstrate that the proposed method significantly outperforms multiple-branch models and achieves the state-of-the-art results on multiple object detection benchmarks (mAP: 49.1% on COCO).

The state-of-the-art object detection method is complicated with various modules such as backbone, RPN, feature fusion neck and RCNN head, where each module may have different designs and structures. How to leverage the computational cost and accuracy trade-off for the structural combination as well as the modular selection of multiple modules? Neural architecture search (NAS) has shown great potential in finding an optimal solution. Existing NAS works for object detection only focus on searching better design of a single module such as backbone or feature fusion neck, while neglecting the balance of the whole system. In this paper, we present a two-stage coarse-to-fine searching strategy named Structural-to-Modular NAS (SM-NAS) for searching a GPU-friendly design of both an efficient combination of modules and better modular-level architecture for object detection. Specifically, Structural-level searching stage first aims to find an efficient combination of different modules; Modular-level searching stage then evolves each specific module and pushes the Pareto front forward to a faster task-specific network. We consider a multi-objective search where the search space covers many popular designs of detection methods. We directly search a detection backbone without pre-trained models or any proxy task by exploring a fast training from scratch strategy. The resulting architectures dominate state-of-the-art object detection systems in both inference time and accuracy and demonstrate the effectiveness on multiple detection datasets, e.g. halving the inference time with additional 1% mAP improvement compared to FPN and reaching 46% mAP with the similar inference time of MaskRCNN.

Vision-dialog navigation posed as a new holy-grail task in vision-language disciplinary targets at learning an agent endowed with the capability of constant conversation for help with natural language and navigating according to human responses. Besides the common challenges faced in visual language navigation, vision-dialog navigation also requires to handle well with the language intentions of a series of questions about the temporal context from dialogue history and co-reasoning both dialogs and visual scenes. In this paper, we propose the Cross-modal Memory Network (CMN) for remembering and understanding the rich information relevant to historical navigation actions. Our CMN consists of two memory modules, the language memory module (L-mem) and the visual memory module (V-mem). Specifically, L-mem learns latent relationships between the current language interaction and a dialog history by employing a multi-head attention mechanism. V-mem learns to associate the current visual views and the cross-modal memory about the previous navigation actions. The cross-modal memory is generated via a vision-to-language attention and a language-to-vision attention. Benefiting from the collaborative learning of the L-mem and the V-mem, our CMN is able to explore the memory about the decision making of historical navigation actions which is for the current step. Experiments on the CVDN dataset show that our CMN outperforms the previous state-of-the-art model by a significant margin on both seen and unseen environments.

Interactive fashion image manipulation, which enables users to edit images with sketches and color strokes, is an interesting research problem with great application value. Existing works often treat it as a general inpainting task and do not fully leverage the semantic structural information in fashion images. Moreover, they directly utilize conventional convolution and normalization layers to restore the incomplete image, which tends to wash away the sketch and color information. In this paper, we propose a novel Fashion Editing Generative Adversarial Network (FE-GAN), which is capable of manipulating fashion images by free-form sketches and sparse color strokes. FE-GAN consists of two modules: 1) a free-form parsing network that learns to control the human parsing generation by manipulating sketch and color; 2) a parsing-aware inpainting network that renders detailed textures with semantic guidance from the human parsing map. A new attention normalization layer is further applied at multiple scales in the decoder of the inpainting network to enhance the quality of the synthesized image. Extensive experiments on high-resolution fashion image datasets demonstrate that the proposed FE-GAN significantly outperforms the state-of-the-art methods on fashion image manipulation.

Advanced object detectors usually adopt a backbone network designed and pretrained by ImageNet classification. Recently neural architecture search (NAS) has emerged to automatically design a task-specific backbone to bridge the gap between the tasks of classification and detection. In this paper, we propose a two-phase serial-to-parallel architecture search framework named SP-NAS towards a flexible task-oriented detection backbone. Specifically, the serial-searching round aims at finding a sequence of serial blocks with optimal scale and output channels in the feature hierarchy by a Swap-Expand-Reignite search algorithm; the parallel-searching phase then assembles several sub-architectures along with the previous searched backbone into a more powerful parallel-structured backbone. We efficiently search a detection backbone by exploring a network morphism strategy on multiple detection benchmarks. The resulting architectures achieve SOTA results, i.e. top performance (LAMR: 0.055) on the automotive detection leaderboard of EuroCityPersons benchmark, improving 2.3% mAP with less FLOPS than NAS-FPN on COCO, and reaching 84.1% AP50 on VOC better than DetNAS and Auto-FPN in terms of both accuracy and speed.

Neural Architecture Search (NAS), aiming at automatically designing network architectures by machines, is expected to bring about a new revolution in machine learning. Despite these high expectation, the effectiveness and efficiency of existing NAS solutions are unclear, with some recent works going so far as to suggest that many existing NAS solutions are no better than random architecture selection. The ineffectiveness of NAS solutions may be attributed to inaccurate architecture evaluation. Specifically, to speed up NAS, recent works have proposed under-training different candidate architectures in a large search space concurrently by using shared network parameters; however, this has resulted in incorrect architecture ratings and furthered the ineffectiveness of NAS. In this work, we propose to modularize the large search space of NAS into blocks to ensure that the potential candidate architectures are fully trained; this reduces the representation shift caused by the shared parameters and leads to the correct rating of the candidates. Thanks to the block-wise search, we can also evaluate all of the candidate architectures within each block. Moreover, we find that the knowledge of a network model lies not only in the network parameters but also in the network architecture. Therefore, we propose to distill the neural architecture (DNA) knowledge from a teacher model to supervise our block-wise architecture search, which significantly improves the effectiveness of NAS. Remarkably, the performance of our searched architectures has exceeded the teacher model, demonstrating the practicability of our method. Finally, our method achieves a state-of-the-art 78.4% top-1 accuracy on ImageNet in a mobile setting. All of our searched models along with the evaluation code are available at https://github.com/changlin31/DNA.

Recent researches on panoptic segmentation resort to a single end-to-end network to combine the tasks of instance segmentation and semantic segmentation. However, prior models only unified the two related tasks at the architectural level via a multi-branch scheme or revealed the underlying correlation between them by unidirectional feature fusion, which disregards the explicit semantic and co-occurrence relations among objects and background. Inspired by the fact that context information is critical to recognize and localize the objects, and inclusive object details are significant to parse the background scene, we thus investigate on explicitly modeling the correlations between object and background to achieve a holistic understanding of an image in the panoptic segmentation task. We introduce a Bidirectional Graph Reasoning Network (BGRNet), which incorporates graph structure into the conventional panoptic segmentation network to mine the intra-modular and inter-modular relations within and between foreground things and background stuff classes. In particular, BGRNet first constructs image-specific graphs in both instance and semantic segmentation branches that enable flexible reasoning at the proposal level and class level, respectively. To establish the correlations between separate branches and fully leverage the complementary relations between things and stuff, we propose a Bidirectional Graph Connection Module to diffuse information across branches in a learnable fashion. Experimental results demonstrate the superiority of our BGRNet that achieves the new state-of-the-art performance on challenging COCO and ADE20K panoptic segmentation benchmarks.

Vision-Language Navigation (VLN) is a task where an agent learns to navigate following a natural language instruction. The key to this task is to perceive both the visual scene and natural language sequentially. Conventional approaches fully exploit vision and language features in cross-modal grounding. However, the VLN task remains challenging, since previous works have implicitly neglected the rich semantic information contained in environments (such as navigation graphs or sub-trajectory semantics). In this paper, we introduce Auxiliary Reasoning Navigation (AuxRN), a framework with four self-supervised auxiliary reasoning tasks to exploit the additional training signals derived from these semantic information. The auxiliary tasks have four reasoning objectives: explaining the previous actions, evaluating the trajectory consistency, estimating the progress and predict the next direction. As a result, these additional training signals help the agent to acquire knowledge of semantic representations in order to reason about its activities and build a thorough perception of environments. Our experiments demonstrate that auxiliary reasoning tasks improve both the performance of the main task and the model generalizability by a large margin. We further demonstrate empirically that an agent trained with self-supervised auxiliary reasoning tasks substantially outperforms the previous state-of-the-art method, being the best existing approach on the standard benchmark.

Automatically evaluating dialogue coherence is a challenging but high-demand ability for developing high-quality open-domain dialogue systems.However, current evaluation metrics consider only surface features or utterancelevel semantics, without explicitly considering the fine-grained topic transition dynamics of dialogue flows.Here, we first consider that the graph structure constituted with topics in a dialogue can accurately depict the underlying communication logic, which is a more natural way to produce persuasive metrics.Capitalized on the topic-level dialogue graph, we propose a new evaluation metric GRADE, which stands for Graph-enhanced Representations for Automatic Dialogue Evaluation.Specifically, GRADE incorporates both coarsegrained utterance-level contextualized representations and fine-grained topic-level graph representations to evaluate dialogue coherence.The graph representations are obtained by reasoning over topic-level dialogue graphs enhanced with the evidence from a commonsense graph, including k-hop neighboring representations and hop-attention weights.Experimental results show that our GRADE significantly outperforms other state-of-the-art metrics on measuring diverse dialogue models in terms of the Pearson and Spearman correlations with human judgements.Besides, we release a new large-scale human evaluation benchmark to facilitate future research on automatic metrics.

A practical automatic textual math word problems (MWPs) solver should be able to solve various textual MWPs while most existing works only focused on one-unknown linear MWPs.Herein, we propose a simple but efficient method called Universal Expression Tree (UET) to make the first attempt to represent the equations of various MWPs uniformly.Then a semantically-aligned universal tree-structured solver (SAU-Solver) based on an encoder-decoder framework is proposed to resolve multiple types of MWPs in a unified model, benefiting from our UET representation.Our SAU-Solver generates a universal expression tree explicitly by deciding which symbol to generate according to the generated symbols' semantic meanings like human solving MWPs.Besides, our SAU-Solver also includes a novel subtree-level semanticallyaligned regularization to further enforce the semantic constraints and rationality of the generated expression tree by aligning with the contextual information.Finally, to validate the universality of our solver and extend the research boundary of MWPs, we introduce a new challenging Hybrid Math Word Problems dataset (HMWP), consisting of three types of MWPs.Experimental results on several MWPs datasets show that our model can solve universal types of MWPs and outperforms several state-of-the-art models 1 .

The topic of speech enhancement has been largely improved recently, especially with the development of generative adversarial networks (GANs). However prior methods simply follow the GAN architectures from computer vision tasks without specific designs for the speech enhancement according to the audio characteristics (i.e., different granularity context), which may leave noise points in some segments or disturb the contents of the original audio. In this work, we make the first attempt to explore the global and local speech features for coarse-to-fine speech enhancement and introduce a Context Pyramid Generative Adversarial Network (CPGAN), which contains a densely-connected feature pyramid generator and a dynamic context granularity discriminator to better eliminate audio noise hierarchically. Extensive experiments demonstrate that our CP-GAN effectively achieves state-of-the-art speech enhancement results and boosts the performance of more high-level speech tasks including automatic speech recognition and speaker recognition.

Similar to language, understanding an image can be considered as a hierarchical decomposition process from scenes to objects, parts, pixels, and the corresponding spatial/contextual relations. However, the existing convolutional networks concentrate on stacking redundant convolutional layers with a large number of kernels in a hierarchical organization to implicitly approximate this decomposition. This may limit the network to learn the semantic information conveyed in the internal feature maps that may reveal minor yet crucial differences for visual understanding. Attempting to tackle this problem, this paper proposes a simple yet effective tree convolution (TreeConv) operation for deep neural networks. Specifically, inspired by the image grammar techniques[73] that serve as a unified framework of object representation, learning, and recognition, our TreeConv designs a generative image grammar, i.e., tree generation rule, to parse the hierarchy of internal feature maps by generating tree structures and implicitly learning the specific visual grammars for each object category. Extensive experiments on a variety of benchmarks, i.e., classification (ImageNet / CIFAR), detection & segmentation (COCO 2017), and person re-identification (CUHK03), demonstrate the superiority of our TreeConv in both boosting the accuracy and reducing the computational cost. The source code will be available at: https://github.com/wanggrun/TreeConv.

Synchronization is a key step in data-parallel distributed machine learning (ML). Different synchronization systems and strategies perform differently, and to achieve optimal parallel training throughput requires synchronization strategies that adapt to model structures and cluster configurations. Existing synchronization systems often only consider a single or a few synchronization aspects, and the burden of deciding the right synchronization strategy is then placed on the ML practitioners, who may lack the required expertise. In this paper, we develop a model- and resource-dependent representation for synchronization, which unifies multiple synchronization aspects ranging from architecture, message partitioning, placement scheme, to communication topology. Based on this representation, we build an end-to-end pipeline, AutoSync, to automatically optimize synchronization strategies given model structures and resource specifications, lowering the bar for data-parallel distributed ML. By learning from low-shot data collected in only 200 trial runs, AutoSync can discover synchronization strategies up to 1.6x better than manually optimized ones. We develop transfer-learning mechanisms to further reduce the auto-optimization cost -- the simulators can transfer among similar model architectures, among similar cluster configurations, or both. We also present a dataset that contains over 10000 synchronization strategies and run-time pairs on a diverse set of models and cluster specifications.

Panoptic segmentation is posed as a new popular test-bed for the state-of-the-art holistic scene understanding methods with the requirement of simultaneously segmenting both foreground things and background stuff. The state-of-the-art panoptic segmentation network exhibits high structural complexity in different network components, i.e. backbone, proposal-based foreground branch, segmentation-based background branch, and feature fusion module across branches, which heavily relies on expert knowledge and tedious trials. In this work, we propose an efficient, cooperative and highly automated framework to simultaneously search for all main components including backbone, segmentation branches, and feature fusion module in a unified panoptic segmentation pipeline based on the prevailing one-shot Network Architecture Search (NAS) paradigm. Notably, we extend the common single-task NAS into the multi-component scenario by taking the advantages of the newly proposed intra-modular search space and problem-oriented inter-modular search space, which helps us to obtain an optimal network architecture that not only performs well in both instance segmentation and semantic segmentation tasks but also be aware of the reciprocal relations between foreground things and background stuff classes. To relieve the vast computation burden incurred by applying NAS to complicated network architectures, we present a novel path-priority greedy search policy to find a robust, transferrable architecture with significantly reduced searching overhead. Our searched architecture, namely Auto-Panoptic, achieves the new state-of-the-art on the challenging COCO and ADE20K benchmarks. Moreover, extensive experiments are conducted to demonstrate the effectiveness of path-priority policy and transferability of Auto-Panoptic across different datasets.

Recently generating natural language explanations has shown very promising results in not only offering interpretable explanations but also providing additional information and supervision for prediction. However, existing approaches usually require a large set of human annotated explanations for training while collecting a large set of explanations is not only time consuming but also expensive. In this paper, we develop a general framework for interpretable natural language understanding that requires only a small set of human annotated explanations for training. Our framework treats natural language explanations as latent variables that model the underlying reasoning process of a neural model. We develop a variational EM framework for optimization where an explanation generation module and an explanation-augmented prediction module are alternatively optimized and mutually enhance each other. Moreover, we further propose an explanation-based self-training method under this framework for semi-supervised learning. It alternates between assigning pseudo-labels to unlabeled data and generating new explanations to iteratively improve each other. Experiments on two natural language understanding tasks demonstrate that our framework can not only make effective predictions in both supervised and semi-supervised settings, but is also able to generate good natural language explanations.

In this paper, we focus on triplet-based deep binary embedding networks for image retrieval task. The triplet loss has been shown to be effective for hashing retrieval. However, most of the triplet-based deep networks treat the triplets equally or select the hard triplets based on the loss. Such strategies do not consider the order relations of the binary codes and ignore the hash encoding when learning the feature representations. To this end, we propose an order-aware reweighting method to effectively train the triplet-based deep networks, which up-weights the important triplets and down-weights the uninformative triplets via the rank lists of the binary codes. First, we present the order-aware weighting factors to indicate the importance of the triplets, which depend on the rank order of binary codes. Then, we reshape the triplet loss to the squared triplet loss such that the loss function will put more weights on the important triplets. The extensive evaluations on several benchmark datasets show that the proposed method achieves significant performance compared with the state-of-the-art baselines.

Generating long and semantic-coherent reports to describe medical images poses great challenges towards bridging visual and linguistic modalities, incorporating medical domain knowledge, and generating realistic and accurate descriptions. We propose a novel Knowledge-driven Encode, Retrieve, Paraphrase (KERP) approach which reconciles traditional knowledge- and retrieval-based methods with modern learning-based methods for accurate and robust medical report generation. Specifically, KERP decomposes medical report generation into explicit medical abnormality graph learning and subsequent natural language modeling. KERP first employs an Encode module that transforms visual features into a structured abnormality graph by incorporating prior medical knowledge; then a Retrieve module that retrieves text templates based on the detected abnormalities; and lastly, a Paraphrase module that rewrites the templates according to specific cases. The core of KERP is a proposed generic implementation unit—Graph Transformer (GTR) that dynamically transforms high-level semantics between graph-structured data of multiple domains such as knowledge graphs, images and sequences. Experiments show that the proposed approach generates structured and robust reports supported with accurate abnormality description and explainable attentive regions, achieving the state-of-the-art results on two medical report benchmarks, with the best medical abnormality and disease classification accuracy and improved human evaluation performance.

Beyond current conversational chatbots or task-oriented dialogue systems that have attracted increasing attention, we move forward to develop a dialogue system for automatic medical diagnosis that converses with patients to collect additional symptoms beyond their self-reports and automatically makes a diagnosis. Besides the challenges for conversational dialogue systems (e.g. topic transition coherency and question understanding), automatic medical diagnosis further poses more critical requirements for the dialogue rationality in the context of medical knowledge and symptom-disease relations. Existing dialogue systems (Madotto, Wu, and Fung 2018; Wei et al. 2018; Li et al. 2017) mostly rely on datadriven learning and cannot be able to encode extra expert knowledge graph. In this work, we propose an End-to-End Knowledge-routed Relational Dialogue System (KR-DS) that seamlessly incorporates rich medical knowledge graph into the topic transition in dialogue management, and makes it cooperative with natural language understanding and natural language generation. A novel Knowledge-routed Deep Q-network (KR-DQN) is introduced to manage topic transitions, which integrates a relational refinement branch for encoding relations among different symptoms and symptomdisease pairs, and a knowledge-routed graph branch for topic decision-making. Extensive experiments on a public medical dialogue dataset show our KR-DS significantly beats stateof-the-art methods (by more than 8% in diagnosis accuracy). We further show the superiority of our KR-DS on a newly collected medical dialogue system dataset, which is more challenging retaining original self-reports and conversational data between patients and doctors.

Many real-world open-domain conversation applications have specific goals to achieve during open-ended chats, such as recommendation, psychotherapy, education, etc.We study the problem of imposing conversational goals on open-domain chat agents.In particular, we want a conversational system to chat naturally with human and proactively guide the conversation to a designated target subject.The problem is challenging as no public data is available for learning such a target-guided strategy.We propose a structured approach that introduces coarse-grained keywords to control the intended content of system responses.We then attain smooth conversation transition through turn-level supervised learning, and drive the conversation towards the target with discourse-level constraints.We further derive a keyword-augmented conversation dataset for the study.Quantitative and human evaluations show our system can produce meaningful and effective conversations, significantly improving over other approaches 1 .

(Unsupervised) Domain Adaptation (DA) seeks for classifying target instances when solely provided with source labeled and target unlabeled examples for training. Learning domain-invariant features helps to achieve this goal, whereas it underpins unlabeled samples drawn from a single or multiple explicit target domains (Multi-target DA). In this paper, we consider a more realistic transfer scenario: our target domain is comprised of multiple sub-targets implicitly blended with each other so that learners could not identify which sub-target each unlabeled sample belongs to. This Blending-target Domain Adaptation (BTDA) scenario commonly appears in practice and threatens the validities of existing DA algorithms, due to the presence of domain gaps and categorical misalignments among these hidden sub-targets. To reap the transfer performance gains in this new scenario, we propose Adversarial Meta-Adaptation Network (AMEAN). AMEAN entails two adversarial transfer learning processes. The first is a conventional adversarial transfer to bridge our source and mixed target domains. To circumvent the intra-target category misalignment, the second process presents as ``learning to adapt'': It deploys an unsupervised meta-learner receiving target data and their ongoing feature-learning feedbacks, to discover target clusters as our ``meta-sub-target'' domains. This meta-sub-targets auto-design our meta-sub-target adaptation loss, which is capable to progressively eliminate the implicit category mismatching in our mixed target. We evaluate AMEAN and a variety of DA algorithms in three benchmarks under the BTDA setup. Empirical results show that BTDA is a quite challenging transfer setup for most existing DA algorithms, yet AMEAN significantly outperforms these state-of-the-art baselines and effectively restrains the negative transfer effects in BTDA.

Prior highly-tuned human parsing models tend to fit towards each dataset in a specific domain or with discrepant label granularity, and can hardly be adapted to other human parsing tasks without extensive re-training. In this paper, we aim to learn a single universal human parsing model that can tackle all kinds of human parsing needs by unifying label annotations from different domains or at various levels of granularity. This poses many fundamental learning challenges, e.g. discovering underlying semantic structures among different label granularity, performing proper transfer learning across different image domains, and identifying and utilizing label redundancies across related tasks. To address these challenges, we propose a new universal human parsing agent, named “Graphonomy”, which incorporates hierarchical graph transfer learning upon the conventional parsing network to encode the underlying label semantic structures and propagate relevant semantic information. In particular, Graphonomy first learns and propagates compact high-level graph representation among the labels within one dataset via Intra-Graph Reasoning, and then transfers semantic information across multiple datasets via Inter-Graph Transfer. Various graph transfer dependencies (e.g., similarity, linguistic knowledge) between different datasets are analyzed and encoded to enhance graph transfer capability. By distilling universal semantic graph representation to each specific task, Graphonomy is able to predict all levels ofparsing labels in one system without piling up the complexity. Experimental results show Graphonomy effectively achieves the state-of-the-art results on three human parsing benchmarks as well as advantageous universal human parsing performance.

Graph convolutional neural networks have recently shown great potential for the task of zero-shot learning. These models are highly sample efficient as related concepts in the graph structure share statistical strength allowing generalization to new classes when faced with a lack of data. However, multi-layer architectures, which are required to propagate knowledge to distant nodes in the graph, dilute the knowledge by performing extensive Laplacian smoothing at each layer and thereby consequently decrease performance. In order to still enjoy the benefit brought by the graph structure while preventing dilution of knowledge from distant nodes, we propose a Dense Graph Propagation (DGP) module with carefully designed direct links among distant nodes. DGP allows us to exploit the hierarchical graph structure of the knowledge graph through additional connections. These connections are added based on a node's relationship to its ancestors and descendants. A weighting scheme is further used to weigh their contribution depending on the distance to the node to improve information propagation in the graph. Combined with finetuning of the representations in a two-stage training approach our method outperforms state-of-the-art zero-shot learning approaches.

The dominant deep learning approaches use a "one-size-fits-all" paradigm with the hope that underlying characteristics of diverse inputs can be captured via a fixed structure. They also overlook the importance of explicitly modeling feature hierarchy. However, complex real-world tasks often require discovering diverse reasoning paths for different inputs to achieve satisfying predictions, especially for challenging large-scale recognition tasks with complex label relations. In this paper, we treat the structured commonsense knowledge (e.g. concept hierarchy) as the guidance of customizing more powerful and explainable network structures for distinct inputs, leading to dynamic and individualized inference paths. Give an off-the-shelf large network configuration, the proposed Personalized Modular Network (PMN) is learned by selectively activating a sequence of network modules where each of them is designated to recognize particular levels of structured knowledge. Learning semantic configurations and activation of modules to align well with structured knowledge can be regarded as a decision-making procedure, which is solved by a new graph-based reinforcement learning algorithm. Experiments on three semantic segmentation tasks and classification tasks show our PMN can achieve superior performance with the reduced number of network modules while discovering personalized and explainable module configurations for each input.

How to proper encode high-order object relation in the detection system without any external knowledge? How to leverage the information between co-occurrence and locations of objects for better reasoning? These questions are key challenges towards large-scale object detection system that aims to recognize thousands of objects entangled with complex spatial and semantic relationships nowadays. Distilling key relations that may affect object recognition is crucially important since treating each region separately leads to a big performance drop when facing heavy long-tail data distributions and plenty of confusing categories. Recent works try to encode relation by constructing graphs, e.g. using handcraft linguistic knowledge between classes or implicitly learning a fully-connected graph between regions. However, the handcraft linguistic knowledge cannot be individualized for each image due to the semantic gap between linguistic and visual context while the fully-connected graph is inefficient and noisy by incorporating redundant and distracted relations/edges from irrelevant objects and backgrounds. In this work, we introduce a Spatial-aware Graph Relation Network (SGRN) to adaptive discover and incorporate key semantic and spatial relationships for reasoning over each object. Our method considers the relative location layouts and interactions among which can be easily injected into any detection pipelines to boost the performance. Specifically, our SGRN integrates a graph learner module for learning a interpatable sparse graph structure to encode relevant contextual regions and a spatial graph reasoning module with learnable spatial Gaussian kernels to perform graph inference with spatial awareness. Extensive experiments verify the effectiveness of our method, e.g. achieving around 32% improvement on VG(3000 classes) and 28% on ADE in terms of mAP.

In this paper, we address the large-scale object detection problem with thousands of categories, which poses severe challenges due to long-tail data distributions, heavy occlusions, and class ambiguities. However, the dominant object detection paradigm is limited by treating each object region separately without considering crucial semantic dependencies among objects. In this work, we introduce a novel Reasoning-RCNN to endow any detection networks the capability of adaptive global reasoning over all object regions by exploiting diverse human commonsense knowledge. Instead of only propagating the visual features on the image directly, we evolve the high-level semantic representations of all categories globally to avoid distracted or poor visual features in the image. Specifically, built on feature representations of basic detection network, the proposed network first generates a global semantic pool by collecting the weights of previous classification layer for each category, and then adaptively enhances each object features via attending different semantic contexts in the global semantic pool. Rather than propagating information from all semantic information that may be noisy, our adaptive global reasoning automatically discovers most relative categories for feature evolving. Our Reasoning-RCNN is light-weight and flexible enough to enhance any detection backbone networks, and extensible for integrating any knowledge resources. Solid experiments on object detection benchmarks show the superiority of our Reasoning-RCNN, e.g. achieving around 16% improvement on VisualGenome, 37% on ADE in terms of mAP and 15% improvement on COCO.

Detecting dense landmarks for diverse clothes, as a fundamental technique for clothes analysis, has attracted increasing research attention due to its huge application potential. However, due to the lack of modeling underlying semantic layout constraints among landmarks, prior works often detect ambiguous and structure-inconsistent landmarks of multiple overlapped clothes in one person. In this paper, we propose to seamlessly enforce structural layout relationships among landmarks on the intermediate representations via multiple stacked layout-graph reasoning layers. We define the layout-graph as a hierarchical structure including a root node, body-part nodes (e.g. upper body, lower body), coarse clothes-part nodes (e.g. collar, sleeve) and leaf landmark nodes (e.g. left-collar, right-collar). Each Layout-Graph Reasoning(LGR) layer aims to map feature representations into structural graph nodes via a Map-to-Node module, performs reasoning over structural graph nodes to achieve global layout coherency via a layout-graph reasoning module, and then maps graph nodes back to enhance feature representations via a Node-to-Map module. The layout-graph reasoning module integrates a graph clustering operation to generate representations of intermediate nodes (bottom-up inference) and then a graph deconvolution operation (top-down inference) over the whole graph. Extensive experiments on two public fashion landmark datasets demonstrate the superiority of our model. Furthermore, to advance the fine-grained fashion landmark research for supporting more comprehensive clothes generation and attribute recognition, we contribute the first Fine-grained Fashion Landmark Dataset (FFLD) containing 200k images annotated with at most 32 key-points for 13 clothes types.

Beyond current image-based virtual try-on systems that have attracted increasing attention, we move a step forward to developing a video virtual try-on system that precisely transfers clothes onto the person and generates visually realistic videos conditioned on arbitrary poses. Besides the challenges in image-based virtual try-on (e.g., clothes fidelity, image synthesis), video virtual try-on further requires spatiotemporal consistency. Directly adopting existing image-based approaches often fails to generate coherent video with natural and realistic textures. In this work, we propose Flow-navigated Warping Generative Adversarial Network (FW-GAN), a novel framework that learns to synthesize the video of virtual try-on based on a person image, the desired clothes image, and a series of target poses. FW-GAN aims to synthesize the coherent and natural video while manipulating the pose and clothes. It consists of: (i) a flow-guided fusion module that warps the past frames to assist synthesis, which is also adopted in the discriminator to help enhance the coherence and quality of the synthesized video; (ii) a warping net that is designed to warp clothes image for the refinement of clothes textures; (iii) a parsing constraint loss that alleviates the problem caused by the misalignment of segmentation maps from images with different poses and various clothes. Experiments on our newly collected dataset show that FW-GAN can synthesize high-quality video of virtual try-on and significantly outperforms other methods both qualitatively and quantitatively.

Virtual try-on systems under arbitrary human poses have significant application potential, yet also raise extensive challenges, such as self-occlusions, heavy misalignment among different poses, and complex clothes textures. Existing virtual try-on methods can only transfer clothes given a fixed human pose, and still show unsatisfactory performances, often failing to preserve person identity or texture details, and with limited pose diversity. This paper makes the first attempt towards a multi-pose guided virtual try-on system, which enables clothes to transfer onto a person with diverse poses. Given an input person image, a desired clothes image, and a desired pose, the proposed Multi-pose Guided Virtual Try-On Network (MG-VTON) generates a new person image after fitting the desired clothes into the person and manipulating the pose. MG-VTON is constructed with three stages: 1) a conditional human parsing network is proposed that matches both the desired pose and the desired clothes shape; 2) a deep Warping Generative Adversarial Network (Warp-GAN) that warps the desired clothes appearance into the synthesized human parsing map and alleviates the misalignment problem between the input human pose and the desired one; 3) a refinement render network recovers the texture details of clothes and removes artifacts, based on multi-pose composition masks. Extensive experiments on commonly-used datasets and our newly-collected largest virtual try-on benchmark demonstrate that our MG-VTON significantly outperforms all state-of-the-art methods both qualitatively and quantitatively, showing promising virtual try-on performances.

Neural architecture search (NAS) has shown great potential in automating the manual process of designing a good CNN architecture for image classification. In this paper, we study NAS for object detection, a core computer vision task that classifies and localizes object instances in an image. Existing works focus on transferring the searched architecture from classification task (ImageNet) to the detector backbone, while the rest of the architecture of the detector remains unchanged. However, this pipeline is not task-specific or data-oriented network search which cannot guarantee optimal adaptation to any dataset. Therefore, we propose an architecture search framework named Auto-FPN specifically designed for detection beyond simply searching a classification backbone. Specifically, we propose two auto search modules for detection: Auto-fusion to search a better fusion of the multi-level features; Auto-head to search a better structure for classification and bounding-box(bbox) regression. Instead of searching for one repeatable cell structure, we relax the constraint and allow different cells. The search space of both modules covers many popular designs of detectors and allows efficient gradient-based architecture search with resource constraint (2 days for COCO on 8 GPU cards). Extensive experiments on Pascal VOC, COCO, BDD, VisualGenome and ADE demonstrate the effectiveness of the proposed method, e.g. achieving around 5% improvement than FPN in terms of mAP while requiring around 50% fewer parameters on the searched modules.

Resembling the rapid learning capability of human, low-shot learning empowers vision systems to understand new concepts by training with few samples. Leading approaches derived from meta-learning on images with a single visual object. Obfuscated by a complex background and multiple objects in one image, they are hard to promote the research of low-shot object detection/segmentation. In this work, we present aflexible and general methodology to achieve these tasks. Our work extends Faster /Mask R-CNN by proposing meta-learning over RoI (Region-of-Interest) features instead of a full image feature. This simple spirit disentangles multi-object information merged with the background, without bells and whistles, enabling Faster /Mask R-CNN turn into a meta-learner to achieve the tasks. Specifically, we introduce a Predictor-head Remodeling Network (PRN) that shares its main backbone with Faster /Mask R-CNN. PRN receives images containing low-shot objects with their bounding boxes or masks to infer their class attentive vectors. The vectors take channel-wise soft-attention on RoI features, remodeling those R-CNN predictor heads to detect or segment the objects consistent with the classes these vectors represent. In our experiments, Meta R-CNN yields the new state of the art in low-shot object detection and improves low-shot object segmentation byMaskR-CNN.Code: https://yanxp.github.io/metarcnn.html.

Manipulating person images under diverse poses, which transfers a person from one pose to another desired pose, is an interesting yet challenging task due to large non-rigid spatial deformation. Most existing works fail to preserve the fine-grained appearance consistency along with the pose changes due to the lack of explicit constraints and spatial modeling, leading to unrealistic results with severe artifacts. In this paper, we propose a novel Part-Preserving Generative Adversarial Network (PP-GAN) to achieve good manipulation quality by explicitly enforcing rich structure constraints over generative modeling. PP-GAN is proposed to decompose the challenging spatial transformation of the whole body into fine-grained part-level transformations, which are then integrated via human joint structure constraint. Given arbitrary poses, PP-GAN integrates human joint structure and region-level part cues as inputs to perform explicit generative modeling. Besides, we introduce a parsing-consistent loss to enforce semantic consistency among images with diverse poses, which guides the image synthesis from a semantic perspective. Extensive qualitative and quantitative evaluations on two benchmarks show that our PP-GAN significantly outperforms the state-of-the-art baselines in generating more realistic and plausible image synthesis results. PP-GAN successfully preserves part-level characteristics even for most challenging pose changes while prior works are easy to fail.

A broad range of cross-$m$-domain generation researches boil down to matching a joint distribution by deep generative models (DGMs). Hitherto algorithms excel in pairwise domains while as $m$ increases, remain struggling to scale themselves to fit a joint distribution. In this paper, we propose a domain-scalable DGM, i.e., MMI-ALI for $m$-domain joint distribution matching. As an $m$-domain ensemble model of ALIs (Dumoulin et al., 2016), MMI-ALI is adversarially trained with maximizing Multivariate Mutual Information (MMI) w.r.t. joint variables of each pair of domains and their shared feature. The negative MMIs are upper bounded by a series of feasible losses provably leading to matching $m$-domain joint distributions. MMI-ALI linearly scales as $m$ increases and thus, strikes a right balance between efficacy and scalability. We evaluate MMI-ALI in diverse challenging $m$-domain scenarios and verify its superiority.

Visual commonsense reasoning task aims at leading the research field into solving cognition-level reasoning with the ability to predict correct answers and meanwhile providing convincing reasoning paths, resulting in three sub-tasks i.e., Q->A, QA->R and Q->AR. It poses great challenges over the proper semantic alignment between vision and linguistic domains and knowledge reasoning to generate persuasive reasoning paths. Existing works either resort to a powerful end-to-end network that cannot produce interpretable reasoning paths or solely explore intra-relationship of visual objects (homogeneous graph) while ignoring the cross-domain semantic alignment among visual concepts and linguistic words. In this paper, we propose a new Heterogeneous Graph Learning (HGL) framework for seamlessly integrating the intra-graph and inter-graph reasoning in order to bridge the vision and language domain. Our HGL consists of a primal vision-to-answer heterogeneous graph (VAHG) module and a dual question-to-answer heterogeneous graph (QAHG) module to interactively refine reasoning paths for semantic agreement. Moreover, our HGL integrates a contextual voting module to exploit a long-range visual context for better global reasoning. Experiments on the large-scale Visual Commonsense Reasoning benchmark demonstrate the superior performance of our proposed modules on three tasks (improving 5% accuracy on Q->A, 3.5% on QA->R, 5.8% on Q->AR).

Detecting all visual relationships is posed as the most fundamental task towards the ultimate semantic reasoning. However, due to the rich context embedded in the image and diverse language ambiguities, it is unrealistic to annotate and list all possible relationships for providing a noise-free supervised setting. All prior approaches simply adopt the traditional fully-supervised detection pipeline and ignore the effect of incomplete annotations on model convergence, resulting in the unstable optimization and unsatisfactory performance. In this work, we make the first attempt to address this critical incomplete annotations issue and reformulate this task via the Soft Transfer Learning (STL), which aims to transfer knowledge learned from the annotations in hand into the uncertain pairs in a self-supervised way. The knowledge transfer process is inferred from a principled gradient diagnosis. Extensive experiments on VRD and the large-scale VG benchmarks demonstrate the superiority of our STL method.

Human parsing and pose estimation have recently received considerable interest due to their substantial application potentials. However, the existing datasets have limited numbers of images and annotations and lack a variety of human appearances and coverage of challenging cases in unconstrained environments. In this paper, we introduce a new benchmark named "Look into Person (LIP)" that provides a significant advancement in terms of scalability, diversity, and difficulty, which are crucial for future developments in human-centric analysis. This comprehensive dataset contains over 50,000 elaborately annotated images with 19 semantic part labels and 16 body joints, which are captured from a broad range of viewpoints, occlusions, and background complexities. Using these rich annotations, we perform detailed analyses of the leading human parsing and pose estimation approaches, thereby obtaining insights into the successes and failures of these methods. To further explore and take advantage of the semantic correlation of these two tasks, we propose a novel joint human parsing and pose estimation network to explore efficient context modeling, which can simultaneously predict parsing and pose with extremely high quality. Furthermore, we simplify the network to solve human parsing by exploring a novel self-supervised structure-sensitive learning approach, which imposes human pose structures into the parsing results without resorting to extra supervision. The datasets, code and models are available at http://www.sysu-hcp.net/lip/.

Salient segmentation aims to segment out attentiongrabbing regions, a critical yet challenging task and the foundation of many high-level computer vision applications. It requires semantic-aware grouping of pixels into salient regions and benefits from the utilization of global multi-scale contexts to achieve good local reasoning. Previous works often address it as two-class segmentation problems utilizing complicated multi-step procedures including refinement networks and complex graphical models. We argue that semantic salient segmentation can instead be effectively resolved by reformulating it as a simple yet intuitive pixel-pair based connectivity prediction task. Following the intuition that salient objects can be naturally grouped via semanticaware connectivity between neighboring pixels, we propose a pure Connectivity Net (ConnNet). ConnNet predicts connectivity probabilities of each pixel with its neighboring pixels by leveraging multi-level cascade contexts embedded in the image and long-range pixel relations. We investigate our approach on two tasks, namely salient object segmentation and salient instancelevel segmentation, and illustrate that consistent improvements can be obtained by modeling these tasks as connectivity instead of binary segmentation tasks for a variety of network architectures. We achieve state-of-the-art performance, outperforming or being comparable to existing approaches while reducing inference time due to our less complex approach.

The collaborative reasoning for understanding each image-question pair is very critical but under-explored for an interpretable Visual Question Answering (VQA) system. Although very recent works also tried the explicit compositional processes to assemble multiple sub-tasks embedded in the questions, their models heavily rely on the annotations or hand-crafted rules to obtain valid reasoning layout, leading to either heavy labor or poor performance on composition reasoning. In this paper, to enable global context reasoning for better aligning image and language domains in diverse and unrestricted cases, we propose a novel reasoning network called Adversarial Composition Modular Network (ACMN). This network comprises of two collaborative modules: i) an adversarial attention module to exploit the local visual evidence for each word parsed from the question; ii) a residual composition module to compose the previously mined evidence. Given a dependency parse tree for each question, the adversarial attention module progressively discovers salient regions of one word by densely combining regions of child word nodes in an adversarial manner. Then residual composition module merges the hidden representations of an arbitrary number of children through sum pooling and residual connection. Our ACMN is thus capable of building an interpretable VQA system that gradually dives the image cues following a question-driven reasoning route and makes global reasoning by incorporating the learned knowledge of all attention modules in a principled manner. Experiments on relational datasets demonstrate the superiority of our ACMN and visualization results show the explainable capability of our reasoning system.

Video object segmentation is a fundamental yet challenging task in computer vision community. In this paper, we formulate this problem as a Markov Decision Process, where agents are learned to segment object regions under a deep reinforcement learning framework. Essentially, learning agents for segmentation is nontrivial as segmentation is a nearly continuous decision-making process, where the number of the involved agents (pixels or superpixels) and action steps from the seed (super)pixels to the whole object mask might be incredibly huge. To overcome this difficulty, this paper simplifies the learning of segmentation agents to the learning of a cutting-agent, which only has a limited number of action units and can converge in just a few action steps. The basic assumption is that object segmentation mainly relies on the interaction between object regions and their context. Thus, with an optimal object (box) region and context (box) region, we can obtain the desirable segmentation mask through further inference. Based on this assumption, we establish a novel reinforcement cutting-agent learning framework, where the cutting-agent consists of a cutting-policy network and a cutting-execution network. The former learns policies for deciding optimal object-context box pair, while the latter executes the cutting function based on the inferred object-context box pair. With the collaborative interaction between the two networks, our method can achieve the outperforming VOS performance on two public benchmarks, which demonstrates the rationality of our assumption as well as the effectiveness of the proposed learning framework.

Semantic concept hierarchy is still under-explored for semantic segmentation due to the inefficiency and complicated optimization of incorporating structural inference into dense prediction. This lack of modeling semantic correlations also makes prior works must tune highly-specified models for each task due to the label discrepancy across datasets. It severely limits the generalization capability of segmentation models for open set concept vocabulary and annotation utilization. In this paper, we propose a Dynamic-Structured Semantic Propagation Network (DSSPN) that builds a semantic neuron graph by explicitly incorporating the semantic concept hierarchy into network construction. Each neuron represents the instantiated module for recognizing a specific type of entity such as a super-class (e.g. food) or a specific concept (e.g. pizza). During training, DSSPN performs the dynamic-structured neuron computation graph by only activating a sub-graph of neurons for each image in a principled way. A dense semantic-enhanced neural block is proposed to propagate the learned knowledge of all ancestor neurons into each fine-grained child neuron for feature evolving. Another merit of such semantic explainable structure is the ability of learning a unified model concurrently on diverse datasets by selectively activating different neuron sub-graphs for each annotation at each step. Extensive experiments on four public semantic segmentation datasets (i.e. ADE20K, COCO-Stuff, Cityscape and Mapillary) demonstrate the superiority of our DSSPN over state-of-the-art segmentation models. Moreoever, we demonstrate a universal segmentation model that is jointly trained on diverse datasets can surpass the performance of the common fine-tuning scheme for exploiting multiple domain knowledge.

Cellular Electron Cryo-Tomography (CECT) is a powerful imaging technique for the 3D visualization of cellular structure and organization at submolecular resolution. It enables analyzing the native structures of macromolecular complexes and their spatial organization inside single cells. However, due to the high degree of structural complexity and practical imaging limitations, systematic macromolecular structural recovery inside CECT images remains challenging. Particularly, the recovery of a macromolecule is likely to be biased by its neighbor structures due to the high molecular crowding. To reduce the bias, here we introduce a novel 3D convolutional neural network inspired by Fully Convolutional Network and Encoder-Decoder Architecture for the supervised segmentation of macromolecules of interest in subtomograms. The tests of our models on realistically simulated CECT data demonstrate that our new approach has significantly improved segmentation performance compared to our baseline approach. Also, we demonstrate that the proposed model has generalization ability to segment new structures that do not exist in training data.

Teaching a robot to predict and mimic how a human moves or acts in the near future by observing a series of historical human movements is a crucial first step in human-robot interaction and collaboration. In this paper, we instrument a robot with such a prediction ability by leveraging recent deep learning and computer vision techniques. First, our system takes images from the robot camera as input to produce the corresponding human skeleton based on real-time human pose estimation obtained with the OpenPose library. Then, conditioning on this historical sequence, the robot forecasts plausible motion through a motion predictor, generating a corresponding demonstration. Because of a lack of high-level fidelity validation, existing forecasting algorithms suffer from error accumulation and inaccurate prediction. Inspired by generative adversarial networks (GANs), we introduce a global discriminator that examines whether the predicted sequence is smooth and realistic. Our resulting motion GAN model achieves superior prediction performance to state-of-the-art approaches when evaluated on the standard H3.6M dataset. Based on this motion GAN model, the robot demonstrates its ability to replay the predicted motion in a human-like manner when interacting with a person.

A mobility event occurs when a passenger moves out or takes off from a particular location. Mobility event prediction is of utmost importance in the field of intelligent transportation systems. It has a huge potential in solving important problems such as minimizing passenger waiting time and maximizing the utilization of the transportation resources by planning vehicle routes and dispatching transportation resources. Recently, numerous mobility pattern mining methods have been proposed to predict the transportation supply and demand in different locations. Those methods first reveal the event patterns of each Place of Interests (POI) independently and then employ a separate region function as a post-processing step. This separate process, that disregards the intrinsic spatial and temporal pattern correlations between POI, is sub-optimal and complex, resulting in a poor generalization in different scenarios. In this work, we propose a Spatial-Temporal mobility Event Prediction framework based on Deep neural network (StepDeep) for simultaneously taking into account all correlated spatial and temporal mobility patterns. StepDeep not only simplifies the prediction process but also enhances the prediction accuracy. Our StepDeep proposes a novel problem formulation towards an end-to-end mobility prediction framework, that is, switching mobility events over time in an area into an event video and then posing the mobility prediction problem as a video prediction task. Such a novel formulation can naturally encode spatial and temporal dependencies for each POI. StepDeep thus predicts the spatial-temporal events by incorporating the new time sensitive convolution filters, spatial sensitive convolution filters, and spatial-temporal sensitive convolution filters into a single network. We conduct experimental evaluations on a real-world 547-day New York City taxi trajectory dataset, which show that StepDeep provides higher prediction accuracy than five existing baselines. Moreover, StepDeep is generalizable and can be applied to numerous spatial-temporal event prediction scenarios.

This article summarizes the corresponding half-day tutorial at ACM Multimedia 2018. This tutorial reviews recent progresses for pixel-level understanding with structured deep learning, including 1) human-centric analysis: human parsing and pose estimation; 2) part-based analysis: object part and face parsing; 3) weakly-supervised analysis: object localization and semantic segmentation; 4) depth estimation: stereo matching.

Beyond the existing single-person and multiple-person human parsing tasks in static images, this paper makes the first attempt to investigate a more realistic video instance-level human parsing that simultaneously segments out each person instance and parses each instance into more fine-grained parts (\eg, head, leg, dress). We introduce a novel Adaptive Temporal Encoding Network (ATEN) that alternatively performs temporal encoding among key frames and flow-guided feature propagation from other consecutive frames between two key frames. Specifically, ATEN first incorporates a Parsing-RCNN to produce the instance-level parsing result for each key frame, which integrates both the global human parsing and instance-level human segmentation into a unified model. To balance between accuracy and efficiency, the flow-guided feature propagation is used to directly parse consecutive frames according to their identified temporal consistency with key frames. On the other hand, ATEN leverages the convolution gated recurrent units (convGRU) to exploit temporal changes over a series of key frames, which are further used to facilitate the frame-level instance-level parsing. By alternatively performing direct feature propagation between consistent frames and temporal encoding network among key frames, our ATEN achieves a good balance between frame-level accuracy and time efficiency, which is a common crucial problem in video object segmentation research. To demonstrate the superiority of our ATEN, extensive experiments are conducted on the most popular video segmentation benchmark (DAVIS) and a newly collected Video Instance-level Parsing (VIP) dataset, which is the first video instance-level human parsing dataset comprised of 404 sequences and over 20k frames with instance-level and pixel-wise annotations.

Despite remarkable advances in image synthesis research, existing works often fail in manipulating images under the context of large geometric transformations. Synthesizing person images conditioned on arbitrary poses is one of the most representative examples where the generation quality largely relies on the capability of identifying and modeling arbitrary transformations on different body parts. Current generative models are often built on local convolutions and overlook the key challenges (e.g. heavy occlusions, different views or dramatic appearance changes) when distinct geometric changes happen for each part, caused by arbitrary pose manipulations. This paper aims to resolve these challenges induced by geometric variability and spatial displacements via a new Soft-Gated Warping Generative Adversarial Network (Warping-GAN), which is composed of two stages: 1) it first synthesizes a target part segmentation map given a target pose, which depicts the region-level spatial layouts for guiding image synthesis with higher-level structure constraints; 2) the Warping-GAN equipped with a soft-gated warping-block learns feature-level mapping to render textures from the original image into the generated segmentation map. Warping-GAN is capable of controlling different transformation degrees given distinct target poses. Moreover, the proposed warping-block is light-weight and flexible enough to be injected into any networks. Human perceptual studies and quantitative evaluations demonstrate the superiority of our Warping-GAN that significantly outperforms all existing methods on two large datasets.

The broad set of deep generative models (DGMs) has achieved remarkable advances. However, it is often difficult to incorporate rich structured domain knowledge with the end-to-end DGMs. Posterior regularization (PR) offers a principled framework to impose structured constraints on probabilistic models, but has limited applicability to the diverse DGMs that can lack a Bayesian formulation or even explicit density evaluation. PR also requires constraints to be fully specified {\it a priori}, which is impractical or suboptimal for complex knowledge with learnable uncertain parts. In this paper, we establish mathematical correspondence between PR and reinforcement learning (RL), and, based on the connection, expand PR to learn constraints as the extrinsic reward in RL. The resulting algorithm is model-agnostic to apply to any DGMs, and is flexible to adapt arbitrary constraints with the model jointly. Experiments on human image generation and templated sentence generation show models with learned knowledge constraints by our algorithm greatly improve over base generative models.

Abstract The dominant object detection approaches treat the recognition of each region separately and overlook crucial semantic correlations between objects in one scene. This paradigm leads to substantial performance drop when facing heavy long-tail problems, where very few samples are available for rare classes and plenty of confusing categories exists. We exploit diverse human commonsense knowledge for reasoning over large-scale object categories and reaching semantic coherency within one image. Particularly, we present Hybrid Knowledge Routed Modules (HKRM) that incorporates the reasoning routed by two kinds of knowledge forms: an explicit knowledge module for structured constraints that are summarized with linguistic knowledge (e.g. shared attributes, relationships) about concepts; and an implicit knowledge module that depicts some implicit constraints (e.g. common spatial layouts). By functioning over a region-to-region graph, both modules can be individualized and adapted to coordinate with visual patterns in each image, guided by specific knowledge forms. HKRM are light-weight, general-purpose and extensible by easily incorporating multiple knowledge to endow any detection networks the ability of global semantic reasoning. Experiments on large-scale object detection benchmarks show HKRM obtains around 34.5% improvement on VisualGenome (1000 categories) and 30.4% on ADE in terms of mAP.

Generating long and coherent reports to describe medical images poses challenges to bridging visual patterns with informative human linguistic descriptions. We propose a novel Hybrid Retrieval-Generation Reinforced Agent (HRGR-Agent) which reconciles traditional retrieval-based approaches populated with human prior knowledge, with modern learning-based approaches to achieve structured, robust, and diverse report generation. HRGR-Agent employs a hierarchical decision-making procedure. For each sentence, a high-level retrieval policy module chooses to either retrieve a template sentence from an off-the-shelf template database, or invoke a low-level generation module to generate a new sentence. HRGR-Agent is updated via reinforcement learning, guided by sentence-level and word-level rewards. Experiments show that our approach achieves the state-of-the-art results on two medical report datasets, generating well-balanced structured sentences with robust coverage of heterogeneous medical report contents. In addition, our model achieves the highest detection precision of medical abnormality terminologies, and improved human evaluation performance.

Beyond local convolution networks, we explore how to harness various external human knowledge for endowing the networks with the capability of semantic global reasoning. Rather than using separate graphical models (e.g. CRF) or constraints for modeling broader dependencies, we propose a new Symbolic Graph Reasoning (SGR) layer, which performs reasoning over a group of symbolic nodes whose outputs explicitly represent different properties of each semantic in a prior knowledge graph. To cooperate with local convolutions, each SGR is constituted by three modules: a) a primal local-to-semantic voting module where the features of all symbolic nodes are generated by voting from local representations; b) a graph reasoning module propagates information over knowledge graph to achieve global semantic coherency; c) a dual semantic-to-local mapping module learns new associations of the evolved symbolic nodes with local representations, and accordingly enhances local features. The SGR layer can be injected between any convolution layers and instantiated with distinct prior graphs. Extensive experiments show incorporating SGR significantly improves plain ConvNets on three semantic segmentation tasks and one image classification task. More analyses show the SGR layer learns shared symbolic representations for domains/datasets with the different label set given a universal knowledge graph, demonstrating its superior generalization capability.

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

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

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

Face hallucination is a domain-specific super-resolution problem with the goal to generate high-resolution (HR) faces from low-resolution (LR) input images. In contrast to existing methods that often learn a single patch-to-patch mapping from LR to HR images and are regardless of the contextual interdependency between patches, we propose a novel Attention-aware Face Hallucination (Attention-FH) framework which resorts to deep reinforcement learning for sequentially discovering attended patches and then performing the facial part enhancement by fully exploiting the global interdependency of the image. Specifically, in each time step, the recurrent policy network is proposed to dynamically specify a new attended region by incorporating what happened in the past. The state (i.e., face hallucination result for the whole image) can thus be exploited and updated by the local enhancement network on the selected region. The Attention-FH approach jointly learns the recurrent policy network and local enhancement network through maximizing the long-term reward that reflects the hallucination performance over the whole image. Therefore, our proposed Attention-FH is capable of adaptively personalizing an optimal searching path for each face image according to its own characteristic. Extensive experiments show our approach significantly surpasses the state-of-the-arts on in-the-wild faces with large pose and illumination variations. The state (i.e., face hallucination result for the whole image) can thus be exploited and updated by the local enhancement network on the selected region. The Attention-FH approach jointly learns the recurrent policy network and local enhancement network through maximizing the long-term reward that reflects the hallucination performance over the whole image. Therefore, our proposed Attention-FH is capable of adaptively personalizing an optimal searching path for each face image according to its own characteristic. Extensive experiments show our approach significantly surpasses the state-of-the-arts on in-the-wild faces with large pose and illumination variations.

Human parsing has recently attracted a lot of research interests due to its huge application potentials. However existing datasets have limited number of images and annotations, and lack the variety of human appearances and the coverage of challenging cases in unconstrained environment. In this paper, we introduce a new benchmark Look into Person (LIP) that makes a significant advance in terms of scalability, diversity and difficulty, a contribution that we feel is crucial for future developments in human-centric analysis. This comprehensive dataset contains over 50,000 elaborately annotated images with 19 semantic part labels, which are captured from a wider range of viewpoints, occlusions and background complexity. Given these rich annotations we perform detailed analysis of the leading human parsing approaches, gaining insights into the success and failures of these methods. Furthermore, in contrast to the existing efforts on improving the feature discriminative capability, we solve human parsing by exploring a novel self-supervised structure-sensitive learning approach, which imposes human pose structures into parsing results without resorting to extra supervision (i.e., no need for specifically labeling human joints in model training). Our self-supervised learning framework can be injected into any advanced neural networks to help incorporate rich high-level knowledge regarding human joints from a global perspective and improve the parsing results. Extensive evaluations on our LIP and the public PASCAL-Person-Part dataset demonstrate the superiority of our method.

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

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

Computers still struggle to understand the interdependency of objects in the scene as a whole, e.g., relations between objects or their attributes. Existing methods often ignore global context cues capturing the interactions among different object instances, and can only recognize a handful of types by exhaustively training individual detectors for all possible relationships. To capture such global interdependency, we propose a deep Variation-structured Re-inforcement Learning (VRL) framework to sequentially discover object relationships and attributes in the whole image. First, a directed semantic action graph is built using language priors to provide a rich and compact representation of semantic correlations between object categories, predicates, and attributes. Next, we use a variation-structured traversal over the action graph to construct a small, adaptive action set for each step based on the current state and historical actions. In particular, an ambiguity-aware object mining scheme is used to resolve semantic ambiguity among object categories that the object detector fails to distinguish. We then make sequential predictions using a deep RL framework, incorporating global context cues and semantic embeddings of previously extracted phrases in the state vector. Our experiments on the Visual Relationship Detection (VRD) dataset and the large-scale Visual Genome dataset validate the superiority of VRL, which can achieve significantly better detection results on datasets involving thousands of relationship and attribute types. We also demonstrate that VRL is able to predict unseen types embedded in our action graph by learning correlations on shared graph nodes.

3D Human articulated pose recovery from monocular image sequences is very challenging due to the diverse appearances, viewpoints, occlusions, and also the human 3D pose is inherently ambiguous from the monocular imagery. It is thus critical to exploit rich spatial and temporal long-range dependencies among body joints for accurate 3D pose sequence prediction. Existing approaches usually manually design some elaborate prior terms and human body kinematic constraints for capturing structures, which are often insufficient to exploit all intrinsic structures and not scalable for all scenarios. In contrast, this paper presents a Recurrent 3D Pose Sequence Machine(RPSM) to automatically learn the image-dependent structural constraint and sequence-dependent temporal context by using a multi-stage sequential refinement. At each stage, our RPSM is composed of three modules to predict the 3D pose sequences based on the previously learned 2D pose representations and 3D poses: (i) a 2D pose module extracting the image-dependent pose representations, (ii) a 3D pose recurrent module regressing 3D poses and (iii) a feature adaption module serving as a bridge between module (i) and (ii) to enable the representation transformation from 2D to 3D domain. These three modules are then assembled into a sequential prediction framework to refine the predicted poses with multiple recurrent stages. Extensive evaluations on the Human3.6M dataset and HumanEva-I dataset show that our RPSM outperforms all state-of-the-art approaches for 3D pose estimation.

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

The recently developed variational autoencoders (VAEs) have proved to be an effective confluence of the rich representational power of neural networks with Bayesian methods. However, most work on VAEs use a rather simple prior over the latent variables such as standard normal distribution, thereby restricting its applications to relatively simple phenomena. In this work, we propose hierarchical non-parametric variational autoencoders, which combines tree-structured Bayesian nonparametric priors with VAEs, to enable infinite flexibility of the latent representation space. Both the neural parameters and Bayesian priors are learned jointly using tailored variational inference. The resulting model induces a hierarchical structure of latent semantic concepts underlying the data corpus, and infers accurate representations of data instances. We apply our model in video representation learning. Our method is able to discover highly interpretable activity hierarchies, and obtain improved clustering accuracy and generalization capacity based on the learned rich representations.

A natural image usually conveys rich semantic content and can be viewed from different angles. Existing image description methods are largely restricted by small sets of biased visual paragraph annotations, and fail to cover rich underlying semantics. In this paper, we investigate a semi-supervised paragraph generative framework that is able to synthesize diverse and semantically coherent paragraph descriptions by reasoning over local semantic regions and exploiting linguistic knowledge. The proposed Recurrent Topic-Transition Generative Adversarial Network (RTT-GAN) builds an adversarial framework between a structured paragraph generator and multi-level paragraph discriminators. The paragraph generator generates sentences recurrently by incorporating region-based visual and language attention mechanisms at each step. The quality of generated paragraph sentences is assessed by multi-level adversarial discriminators from two aspects, namely, plausibility at sentence level and topic-transition coherence at paragraph level. The joint adversarial training of RTT-GAN drives the model to generate realistic paragraphs with smooth logical transition between sentence topics. Extensive quantitative experiments on image and video paragraph datasets demonstrate the effectiveness of our RTT-GAN in both supervised and semi-supervised settings. Qualitative results on telling diverse stories for an image verify the interpretability of RTT-GAN.

Future frame prediction in videos is a promising avenue for unsupervised video representation learning. Video frames are naturally generated by the inherent pixel flows from preceding frames based on the appearance and motion dynamics in the video. However, existing methods focus on directly hallucinating pixel values, resulting in blurry predictions. In this paper, we develop a dual motion Generative Adversarial Net (GAN) architecture, which learns to explicitly enforce future-frame predictions to be consistent with the pixel-wise flows in the video through a dual-learning mechanism. The primal future-frame prediction and dual future-flow prediction form a closed loop, generating informative feedback signals to each other for better video prediction. To make both synthesized future frames and flows indistinguishable from reality, a dual adversarial training method is proposed to ensure that the future-flow prediction is able to help infer realistic future-frames, while the future-frame prediction in turn leads to realistic optical flows. Our dual motion GAN also handles natural motion uncertainty in different pixel locations with a new probabilistic motion encoder, which is based on variational autoencoders. Extensive experiments demonstrate that the proposed dual motion GAN significantly outperforms state-of-the-art approaches on synthesizing new video frames and predicting future flows. Our model generalizes well across diverse visual scenes and shows superiority in unsupervised video representation learning.

In this paper, we investigate a weakly-supervised object detection framework. Most existing frameworks focus on using static images to learn object detectors. However, these detectors often fail to generalize to videos because of the existing domain shift. Therefore, we investigate learning these detectors directly from boring videos of daily activities. Instead of using bounding boxes, we explore the use of action descriptions as supervision since they are relatively easy to gather. A common issue, however, is that objects of interest that are not involved in human actions are often absent in global action descriptions known as “missing label”. To tackle this problem, we propose a novel temporal dynamic graph Long Short-Term Memory network (TDGraph LSTM). TD-Graph LSTM enables global temporal reasoning by constructing a dynamic graph that is based on temporal correlations of object proposals and spans the entire video. The missing label issue for each individual frame can thus be significantly alleviated by transferring knowledge across correlated objects proposals in the whole video. Extensive evaluations on a large-scale daily-life action dataset (i.e., Charades) demonstrates the superiority of our proposed method. We also release object bounding-box annotations for more than 5,000 frames in Charades. We believe this annotated data can also benefit other research on video-based object recognition in the future.

Generic generation and manipulation of text is challenging and has limited success compared to recent deep generative modeling in visual domain. This paper aims at generating plausible text sentences, whose attributes are controlled by learning disentangled latent representations with designated semantics. We propose a new neural generative model which combines variational auto-encoders (VAEs) and holistic attribute discriminators for effective imposition of semantic structures. The model can alternatively be seen as enhancing VAEs with the wake-sleep algorithm for leveraging fake samples as extra training data. With differentiable approximation to discrete text samples, explicit constraints on independent attribute controls, and efficient collaborative learning of generator and discriminators, our model learns interpretable representations from even only word annotations, and produces short sentences with desired attributes of sentiment and tenses. Quantitative experiments using trained classifiers as evaluators validate the accuracy of sentence and attribute generation.

Retrieving image content with a natural language expression is an emerging interdisciplinary problem at the intersection of multimedia, natural language processing and artificial intelligence. Existing methods tackle this challenging problem by learning features from the visual and linguistic domains independently while the critical semantic correlations bridging two domains have been under-explored in the feature learning process. In this paper, we propose to exploit sharable semantic attributes as "anchors" to ensure the learned features are well aligned across domains for better object retrieval. We define "attributes" as the common concepts that are informative for object retrieval and can be easily learned from both visual content and language expression. In particular, diverse and complex attributes (e.g., location, color, category, interaction between object and context) are modeled and incorporated to promote cross-domain alignment for feature learning from multiple perspectives. Based on the sharable attributes, we propose a deep Attribute-Preserving Metric learning (AP-Metric) framework that jointly generates unique query-sensitive region proposals and conducts novel cross-modal feature learning that explicitly pursues consistency over semantic attribute abstraction within both domains for deep metric learning. Benefiting from the cross-modal semantic correlations, our proposed framework can localize challenging visual objects to match complex query expressions within cluttered background accurately. The overall framework is end-to-end trainable. Extensive evaluations on popular datasets including ReferItGame, RefCOCO, and RefCOCO+ well demonstrate its superiority. Notably, it achieves state-of-the-art performance on the challenging ReferItGame dataset.

We study the problem of conditional generative modeling based on designated semantics or structures. Existing models that build conditional generators either require massive labeled instances as supervision or are unable to accurately control the semantics of generated samples. We propose structured generative adversarial networks (SGANs) for semi-supervised conditional generative modeling. SGAN assumes the data x is generated conditioned on two independent latent variables: y that encodes the designated semantics, and z that contains other factors of variation. To ensure disentangled semantics in y and z, SGAN builds two collaborative games in the hidden space to minimize the reconstruction error of y and z, respectively. Training SGAN also involves solving two adversarial games that have their equilibrium concentrating at the true joint data distributions p(x, z) and p(x, y), avoiding distributing the probability mass diffusely over data space that MLE-based methods may suffer. We assess SGAN by evaluating its trained networks, and its performance on downstream tasks. We show that SGAN delivers a highly controllable generator, and disentangled representations; it also establishes start-of-the-art results across multiple datasets when applied for semi-supervised image classification (1.27%, 5.73%, 17.26% error rates on MNIST, SVHN and CIFAR-10 using 50, 1000 and 4000 labels, respectively). Benefiting from the separate modeling of y and z, SGAN can generate images with high visual quality and strictly following the designated semantic, and can be extended to a wide spectrum of applications, such as style transfer.

MOTIVATION: Cellular Electron CryoTomography (CECT) enables 3D visualization of cellular organization at near-native state and in sub-molecular resolution, making it a powerful tool for analyzing structures of macromolecular complexes and their spatial organizations inside single cells. However, high degree of structural complexity together with practical imaging limitations makes the systematic de novo discovery of structures within cells challenging. It would likely require averaging and classifying millions of subtomograms potentially containing hundreds of highly heterogeneous structural classes. Although it is no longer difficult to acquire CECT data containing such amount of subtomograms due to advances in data acquisition automation, existing computational approaches have very limited scalability or discrimination ability, making them incapable of processing such amount of data. RESULTS: To complement existing approaches, in this article we propose a new approach for subdividing subtomograms into smaller but relatively homogeneous subsets. The structures in these subsets can then be separately recovered using existing computation intensive methods. Our approach is based on supervised structural feature extraction using deep learning, in combination with unsupervised clustering and reference-free classification. Our experiments show that, compared with existing unsupervised rotation invariant feature and pose-normalization based approaches, our new approach achieves significant improvements in both discrimination ability and scalability. More importantly, our new approach is able to discover new structural classes and recover structures that do not exist in training data. AVAILABILITY AND IMPLEMENTATION: Source code freely available at http://www.cs.cmu.edu/∼mxu1/software . CONTACT: [email protected]. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

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

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

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

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

This paper proposes a novel optimization scheme by hybridizing an artificial bee colony optimizer (HABC) with a bee life-cycle mechanism, for both stationary and dynamic optimization problems. The main innovation of the proposed HABC is to develop a cooperative and population-varying scheme, in which individuals can dynamically shift their states of birth, foraging, death, and reproduction throughout the artificial bee colony life cycle. That is, the bee colony size can be adjusted dynamically according to the local fitness landscape during algorithm execution. This new characteristic of HABC helps to avoid redundant search and maintain diversity of population in complex environments. A comprehensive experimental analysis is implemented that the proposed algorithm is benchmarked against several state-of-the-art bio-inspired algorithms on both stationary and dynamic benchmarks. Then the proposed HABC is applied to the real-world applications including data clustering and image segmentation problems. Statistical analysis of all these tests highlights the significant performance improvement due to the life-cycle mechanism and shows that the proposed HABC outperforms the reference algorithms.

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

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

This paper addresses a fundamental problem of scene understanding: How to parse the scene image into a structured configuration (i.e., a semantic object hierarchy with object interaction relations) that finely accords with human perception. We propose a deep architecture consisting of two networks: i) a convolutional neural network (CNN) extracting the image representation for pixelwise object labeling and ii) a recursive neural network (RNN) discovering the hierarchical object structure and the inter-object relations. Rather than relying on elaborative user annotations (e.g., manually labeling semantic maps and relations), we train our deep model in a weakly-supervised manner by leveraging the descriptive sentences of the training images. Specifically, we decompose each sentence into a semantic tree consisting of nouns and verb phrases, and facilitate these trees discovering the configurations of the training images. Once these scene configurations are determined, then the parameters of both the CNN and RNN are updated accordingly by back propagation. The entire model training is accomplished through an Expectation-Maximization method. Extensive experiments suggest that our model is capable of producing meaningful and structured scene configurations and achieving more favorable scene labeling performance on PASCAL VOC 2012 over other state-of-the-art weakly-supervised methods.

Human pose estimation (i.e., locating the body parts / joints of a person) is a fundamental problem in human-computer interaction and multimedia applications. Significant progress has been made based on the development of depth sensors, i.e., accessible human pose prediction from still depth images~\cite{rf12pami}. However, most of the existing approaches to this problem involve several components/models that are independently designed and optimized, leading to suboptimal performances. In this paper, we propose a novel inference-embedded multi-task learning framework for predicting human pose from still depth images, which is implemented with a deep architecture of neural networks. Specifically, we handle two cascaded tasks: i) generating the heat (confidence) maps of body parts via a fully convolutional network (FCN); ii) seeking the optimal configuration of body parts based on the detected body part proposals via an inference built-in MatchNet~\cite{mn15cvpr}, which measures the appearance and geometric kinematic compatibility of body parts and embodies the dynamic programming inference as an extra network layer. These two tasks are jointly optimized. Our extensive experiments show that the proposed deep model significantly improves the accuracy of human pose estimation over other several state-of-the-art methods or SDKs. We also release a large-scale dataset for comparison, which includes 100K depth images under challenging scenarios.

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

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

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

This work investigates how to automatically classify Focal Liver Lesions (FLLs) into three specific benign or malignant types in Contrast-Enhanced Ultrasound (CEUS) videos, and aims at providing a computational framework to assist clinicians in FLL diagnosis. The main challenge for this task is that FLLs in CEUS videos often show diverse enhancement patterns at different temporal phases. To handle these diverse patterns, we propose a novel structured model, which detects a number of discriminative Regions of Interest (ROIs) for the FLL and recognize the FLL based on these ROIs. Our model incorporates an ensemble of local classifiers in the attempt to identify different enhancement patterns of ROIs, and in particular, we make the model reconfigurable by introducing switch variables to adaptively select appropriate classifiers during inference. We formulate the model learning as a non-convex optimization problem, and present a principled optimization method to solve it in a dynamic manner: the latent structures (e.g. the selections of local classifiers, and the sizes and locations of ROIs) are iteratively determined along with the parameter learning. Given the updated model parameters in each step, the data-driven inference is also proposed to efficiently determine the latent structures by using the sequential pruning and dynamic programming method. In the experiments, we demonstrate superior performances over the state-of-the-art approaches. We also release hundreds of CEUS FLLs videos used to quantitatively evaluate this work, which to the best of our knowledge forms the largest dataset in the literature. Please find more information at "http://vision.sysu.edu.cn/projects/fllrecog/".

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

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

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

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

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

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

In this paper, we explore how to utilize the video context to facilitate fashion parsing. Instead of annotating a large amount of fashion images, we present a general, affordable and scalable solution, which harnesses the rich contexts in easily available fashion videos to boost any existing fashion parser. First, we crawl a large unlabelled fashion video corpus with fashion frames. Then for each fashion video, the cross-frame contexts are utilized for human pose co-estimation, and then video co-parsing to obtain satisfactory fashion parsing results for all frames. More specifically, Sift Flow and super-pixel matching are used to build correspondences across frames, and these correspondences then con- textualize the pose estimations and fashion parsing in individual frames. Finally, these parsed video frames are used as the reference corpus for the non-parametric fashion parsing component of the whole solution. Extensive experiments on two benchmark fashion datasets as well as a newly collected challenging Fashion Icon (FI) dataset demonstrate the encouraging performance gain from our general pipeline for fashion parsing.

Although it has been widely discussed in video surveillance, background subtraction is still an open problem in the context of complex scenarios, e.g., dynamic backgrounds, illumination variations, and indistinct foreground objects. To address these challenges, we propose an effective background subtraction method by learning and maintaining an array of dynamic texture models within the spatio-temporal representations. At any location of the scene, we extract a sequence of regular video bricks, i.e., video volumes spanning over both spatial and temporal domain. The background modeling is thus posed as pursuing subspaces within the video bricks while adapting the scene variations. For each sequence of video bricks, we pursue the subspace by employing the auto regressive moving average model that jointly characterizes the appearance consistency and temporal coherence of the observations. During online processing, we incrementally update the subspaces to cope with disturbances from foreground objects and scene changes. In the experiments, we validate the proposed method in several complex scenarios, and show superior performances over other state-of-the-art approaches of background subtraction. The empirical studies of parameter setting and component analysis are presented as well.

Action recognition is an important problem in multimedia understanding. This paper addresses this problem by building an expressive compositional action model. We model one action instance in the video with an ensemble of spatio-temporal compositions: a number of discrete temporal anchor frames, each of which is further decomposed to a layout of deformable parts. In this way, our model can identify a Spatio-Temporal And-Or Graph (STAOG) to represent the latent structure of actions \emph{e.g.} triple jumping, swinging and high jumping. The STAOG model comprises four layers: (i) a batch of leaf-nodes in bottom for detecting various action parts within video patches; (ii) the or-nodes over bottom, i.e. switch variables to activate their children leaf-nodes for structural variability; (iii) the and-nodes within an anchor frame for verifying spatial composition; and (iv) the root-node at top for aggregating scores over temporal anchor frames. Moreover, the contextual interactions are defined between leaf-nodes in both spatial and temporal domains. For model training, we develop a novel weakly supervised learning algorithm which iteratively determines the structural configuration (e.g. the production of leaf-nodes associated with the or-nodes) along with the optimization of multi-layer parameters. By fully exploiting spatio-temporal compositions and interactions, our approach handles well large intra-class action variance (\emph{e.g.} different views, individual appearances, spatio-temporal structures). The experimental results on the challenging databases demonstrate superior performance of our approach over other methods.