Dual-lens video inpainting aims to simultaneously restore missing or corrupted contents in videos captured by each lens of binocular systems. Although preliminary explorations have been conducted, existing methods still face two key challenges: limited exploitation of long-range reference information and inadequate modeling of inter-lens consistency in non-standard binocular systems. In this paper, we propose a novel dual-lens video inpainting framework named DLVINet, which addresses these challenges with two core components. Firstly, we develop a sparse spatial-temporal transformer (SSTT) that effectively utilizes the information from distant frames to complete the video contents of each lens individually. By employing sparse spatial-temporal attention with a channel selection mechanism, SSTT not only restores missing regions, but also avoids introducing redundant or irrelevant information. Furthermore, SSTT introduces a multi-scale feed-forward network to enrich the multi-scale representation of completed features. Secondly, we design a cross-lens texture transformer (CLTT) to model inter-lens consistency. By interacting with corresponding features between lenses under the guidance of cross-attention, CLTT captures global inter-lens correspondences. Such a design enables effective cross-view information modeling without being constrained by horizontal parallax, which is particularly critical for non-standard binocular systems. Extensive experiments demonstrate the effectiveness of our DLVINet.

Large Language Models (LLMs) often fail to utilize their latent reasoning capabilities due to a distributional mismatch between ambiguous human inquiries and the structured logic required for machine activation.Existing alignment methods either incur prohibitive O(N ) costs by fine-tuning each model individually or rely on static prompts that fail to resolve query-level structural complexity.In this paper, we propose ReQueR (Reinforcement Query Refinement), a modular framework that treats reasoning elicitation as an inference-time alignment task.We train a specialized Refiner policy via Reinforcement Learning to rewrite raw queries into explicit logical decompositions, treating frozen LLMs as the environment.Rooted in the classical Zone of Proximal Development from educational psychology, we introduce the Adaptive Solver Hierarchy, a curriculum mechanism that stabilizes training by dynamically aligning environmental difficulty with the Refiner's evolving competence.ReQueR yields consistent absolute gains of 1.7%-7.2%across diverse architectures and benchmarks, outperforming strong baselines by 2.1% on average.Crucially, it provides a promising paradigm for one-to-many inference-time reasoning elicitation, enabling a single Refiner trained on a small set of models to effectively unlock reasoning in diverse unseen models.

Real-world document question-answering that relies on screenshots, such as bills and forms, requires evidence that is often spatially localised and visually cluttered. However, most Screenshot Language Models (SLMs) encode the entire page holistically and rely on implicit attention to ''find'' relevant content, which limits both accuracy and efficiency. We present RegionSLM, a region-aware SLM designed to explicitly connect the question to its supporting regions. RegionSLM has two key components: (1) a patch-relevance router that learns a query–region relevance distribution, enabling the model to produce a box-free relevance prior at inference; and (2) Relevance-Guided Region Pooling (RGRP), a query-conditional attention–pooling module that aggregates dense features into a small set of region tokens, which preserves grounding signals while reducing computational overhead. To support training and evaluation, we further curate ReDoc, a region-supervised corpus with 105k documents and 350k question-answer pairs, obtained via a question-guided two-step filtering procedure. Extensive experiments on 12 datasets demonstrate that explicitly learning query–region relevance and pooling it into compact region tokens is an effective and practical recipe for document retrieval and understanding.

Linear attention mechanisms have emerged as efficient alternatives to full self-attention in Graph Transformers, offering linear time complexity. However, existing linear attention models often suffer from a significant drop in expressiveness due to low-rank projection structures and overly uniform attention distributions. We theoretically prove that these properties reduce the class separability of node representations, limiting the model's classification ability. To address this, we propose a novel hybrid framework that enhances both the rank and focus of attention. Specifically, we enhance linear attention by attaching a gated local graph network branch to the value matrix, thereby increasing the rank of the resulting attention map. Furthermore, to alleviate the excessive smoothing effect inherent in linear attention, we introduce a learnable log-power function into the attention scores to reduce entropy and sharpen focus. We theoretically show that this function decreases entropy in the attention distribution, enhancing the separability of learned embeddings. Extensive experiments on both homophilic and heterophilic graph benchmarks demonstrate that our method achieves competitive performance while preserving the scalability of linear attention.

Zero-shot learning (ZSL) aims to recognize unseen classes by transferring semantic knowledge from seen classes to unseen ones, guided by semantic information. To this end, existing works have demonstrated remarkable performance by utilizing global visual features from Convolutional Neural Networks (CNNs) or Vision Transformers (ViTs) for visual-semantic interactions. Due to the limited receptive fields of CNNs and the quadratic complexity of ViTs, however, these visual backbones achieve suboptimal visual-semantic interactions. In this paper, motivated by the visual state space model (i.e., Vision Mamba), which is capable of capturing long-range dependencies and modeling complex visual dynamics, we propose a parameter-efficient ZSL framework called ZeroMamba to advance ZSL. Our ZeroMamba comprises three key components: Semantic-aware Local Projection (SLP), Global Representation Learning (GRL), and Semantic Fusion (SeF). Specifically, SLP integrates semantic embeddings to map visual features to local semantic-related representations, while GRL encourages the model to learn global semantic representations. SeF combines these two semantic representations to enhance the discriminability of semantic features. We incorporate these designs into Vision Mamba, forming an end-to-end ZSL framework. As a result, the learned semantic representations are better suited for classification. Through extensive experiments on four prominent ZSL benchmarks, ZeroMamba demonstrates superior performance, significantly outperforming the state-of-the-art (i.e., CNN-based and ViT-based) methods under both conventional ZSL (CZSL) and generalized ZSL (GZSL) settings.

Micro-Action Recognition (MAR) has gained increasing attention due to its crucial role as a form of non-verbal communication in social interactions, with promising potential for applications in human communication and emotion analysis. However, current approaches often overlook the inherent ambiguity in micro-actions, which arises from the wide category range and subtle visual differences between categories. This oversight hampers the accuracy of micro-action recognition. In this paper, we propose a novel Prototypical Calibrating Ambiguous Network (PCAN) to unleash and mitigate the ambiguity of MAR. Firstly, we employ a hierarchical action-tree to identify the ambiguous sample, categorizing them into distinct sets of ambiguous samples of false negatives and false positives, considering both body- and action-level categories. Secondly, we implement an ambiguous contrastive refinement module to calibrate these ambiguous samples by regulating the distance between ambiguous samples and their corresponding prototypes. This calibration process aims to pull false negative (FN) samples closer to their respective prototypes and push false positive (FP) samples apart from their affiliated prototypes. In addition, we propose a new prototypical diversity amplification loss to strengthen the model's capacity by amplifying the differences between different prototypes. Finally, we propose a prototype-guided rectification to rectify prediction by incorporating the representability of prototypes. Extensive experiments conducted on the benchmark dataset demonstrate the superior performance of our method compared to existing approaches.

Diffusion-based Text-to-Image (T2I) models have demonstrated significant potential in image restoration. However, existing models continue to grapple with challenges such as complex training and prompt design. We introduce a new perspective for improving image restoration by injecting knowledge from pretrained vision-language models into current T2I models. We empirically show that the degradation and content representations in BLIP-2 can be linearly separated, providing promising degradation guidance for image restoration. Specifically, the Feature Difference Instruction (FDI) is first extracted by Q-Formers through a simple subtraction operation based on reference image pairs. Then, we propose a multi-scale FDI adapter to decouple the degradation style and corrupted artifacts, and inject the styleflow exclusively into specific blocks through adapter-tuning, thereby preventing noise interference and eschewing the need for cumbersome weight retraining. In this way, we can train various task-specific adapters according to different degradations, achieving rich detail enhancement in the restoration results. Furthermore, the proposed FDI adapters have attractive properties of practical value, such as composability and generalization ability for all-in-one and mixed-degradation restoration. Extensive experiments under various settings demonstrate that our method has promising repairing quality over 10 image restoration tasks and a wide range of other applications.

Diffusion models, particularly latent diffusion models, have demonstrated remarkable success in text-driven human motion generation. However, it remains challenging for latent diffusion models to effectively compose multiple semantic concepts into a single, coherent motion sequence. To address this issue, we propose EnergyMoGen, which includes two spectrums of Energy-Based Models: ❶ We interpret the diffusion model as a latent-aware energy-based model that generates motions by composing a set of diffusion models in latent space; ❷ We introduce a semantic-aware energy model based on cross-attention, which enables semantic composition and adaptive gradient descent for text embeddings. To overcome the challenges of semantic inconsistency and motion distortion across these two spectrums, we introduce Synergistic Energy Fusion. This design allows the motion latent diffusion model to synthesize high-quality, complex motions by combining multiple energy terms corresponding to textual descriptions. Experiments show that our approach outperforms existing state-of-the-art models on various motion generation tasks, including text-to-motion generation, compositional motion generation, and multi-concept motion generation. Additionally, we demonstrate that our method can be used to extend motion datasets and improve the text-to-motion task.

Animatable head avatar generation typically requires extensive data for training. To reduce the data requirements, a natural solution is to leverage existing data-free static avatar generation methods, such as pre-trained diffusion models with score distillation sampling (SDS), which align avatars with pseudo ground-truth outputs from the diffusion model. However, directly distilling 4D avatars from video diffusion often leads to over-smooth results due to spatial and temporal inconsistencies in the generated video. To address this issue, we propose Zero-1-to-A, a robust method that synthesizes a spatial and temporal consistency dataset for 4D avatar reconstruction using the video diffusion model. Specifically, Zero-1-to-A iteratively constructs video datasets and optimizes animatable avatars in a progressive manner, ensuring that avatar quality increases smoothly and consistently throughout the learning process. This progressive learning involves two stages: (1) Spatial Consistency Learning fixes expressions and learns from front-to-side views, and (2) Temporal Consistency Learning fixes views and learns from relaxed to exaggerated expressions, generating 4D avatars in a simple-to-complex manner. Extensive experiments demonstrate that Zero-1-to-A improves fidelity, animation quality, and rendering speed compared to existing diffusion-based methods, providing a solution for lifelike avatar creation. Code is publicly available at: https://github.com/ZhenglinZhou/Zero-1-to-A.

Human body actions are an important form of non-verbal communication in social interactions. This paper specifically focuses on a subset of body actions known as micro-actions, which are subtle, low-intensity body movements with promising applications in human emotion analysis. In real-world scenarios, human micro-actions often temporally co-occur, with multiple micro-actions overlapping in time, such as concurrent head and hand movements. However, current research primarily focuses on recognizing individual micro-actions while overlooking their co-occurring nature. To address this gap, we propose a new task named Multi-label Micro-Action Detection (MMAD), which involves identifying all micro-actions in a given short video, determining their start and end times, and categorizing them. Accomplishing this requires a model capable of accurately capturing both long-term and short-term action relationships to detect multiple overlapping micro-actions. To facilitate the MMAD task, we introduce a new dataset named Multi-label Micro-Action-52 (MMA-52) and propose a baseline method equipped with a dual-path spatial-temporal adapter to address the challenges of subtle visual change in MMAD. We hope that MMA-52 can stimulate research on micro-action analysis in videos and prompt the development of spatio-temporal modeling in human-centric video understanding. The proposed MMA-52 dataset is available at: https://github.com/VUT-HFUT/Micro-Action.

Video inpainting aims to fill in corrupted regions of the video with plausible contents. Existing methods generally assume that the locations of corrupted regions are known, focusing primarily on the "how to inpaint". This reliance necessitates manual annotation of the corrupted regions using binary masks to indicate "whereto inpaint". However, the annotation of these masks is labor-intensive and expensive, limiting the practicality of current methods. In this paper, we expect to relax this assumption by defining a new blind video inpainting setting, enabling the networks to learn the mapping from corrupted video to inpainted result directly, eliminating the need of corrupted region annotations. Specifically, we propose an end-to-end blind video inpainting network (BVINet) to address both "where to inpaint" and "how to inpaint" simultaneously. On the one hand, BVINet can predict the masks of corrupted regions by detecting semantic-discontinuous regions of the frame and utilizing temporal consistency prior of the video. On the other hand, the predicted masks are incorporated into the BVINet, allowing it to capture valid context information from uncorrupted regions to fill in corrupted ones. Besides, we introduce a consistency loss to regularize the training parameters of BVINet. In this way, mask prediction and video completion mutually constrain each other, thereby maximizing the overall performance of the trained model. Furthermore, we customize a dataset consisting of synthetic corrupted videos, real-world corrupted videos, and their corresponding completed videos. This dataset serves as a valuable resource for advancing blind video inpainting research. Extensive experimental results demonstrate the effectiveness and superiority of our method.

We present InfiniDreamer, a novel framework for arbitrarily long human motion generation. InfiniDreamer addresses the limitations of current motion generation methods, which are typically restricted to short sequences due to the lack of long motion training data. To achieve this, we first generate sub-motions corresponding to each textual description and then assemble them into a coarse, extended sequence using randomly initialized transition segments. We then introduce an optimization-based method called Segment Score Distillation (SSD) to refine the entire long motion sequence. SSD is designed to utilize an existing motion prior, which is trained only on short clips, in a training-free manner. Specifically, SSD iteratively refines overlapping short segments sampled from the coarsely extended long motion sequence, progressively aligning them with the pre-trained motion diffusion prior. This process ensures local coherence within each segment, while the refined transitions between segments maintain global consistency across the entire sequence. Extensive qualitative and quantitative experiments validate the superiority of our framework, showcasing its ability to generate coherent, contextually aware motion sequences of arbitrary length.

Zeolites are crystalline porous materials that have been widely utilized in petrochemical industries as well as sustainable chemistry areas. Synthesis of zeolites often requires small molecules termed Organic Structure Directing Agents (OSDAs), which are critical in forming the porous structure. Molecule generation models can aid the design of OSDAs, but they are limited by single functionality and lack of interactivity. Meanwhile, large language models (LLMs) such as GPT-4, as general-purpose artificial intelligence systems, excel in instruction comprehension, logical reasoning, and interactive communication. However, LLMs lack in-depth chemistry knowledge and first-principle computation capabilities, resulting in uncontrollable outcomes even after fine-tuning. In this paper, we propose OSDA Agent, an interactive OSDA design framework that leverages LLMs as the brain, coupled with computational chemistry tools. The OSDA Agent consists of three main components: the Actor, responsible for generating potential OSDA structures; the Evaluator, which assesses and scores the generated OSDAs using computational chemistry tools; and the Self-reflector, which produces reflective summaries based on the Evaluator's feedback to refine the Actor's subsequent outputs. Experiments on representative zeolite frameworks show the generation-evaluation-reflection-refinement workflow can perform de novo design of OSDAs with superior generation quality than the pure LLM model, generating candidates consistent with experimentally validated OSDAs and optimizing known OSDAs.

Recent advancements in diffusion models have significantly improved video generation and editing capabilities. However, multi-grained video editing, which encompasses class-level, instance-level, and part-level modifications, remains a formidable challenge. The major difficulties in multi-grained editing include semantic misalignment of text-to-region control and feature coupling within the diffusion model. To address these difficulties, we present VideoGrain, a zero-shot approach that modulates space-time (cross- and self-) attention mechanisms to achieve fine-grained control over video content. We enhance text-to-region control by amplifying each local prompt's attention to its corresponding spatial-disentangled region while minimizing interactions with irrelevant areas in cross-attention. Additionally, we improve feature separation by increasing intra-region awareness and reducing inter-region interference in self-attention. Extensive experiments demonstrate our method achieves state-of-the-art performance in real-world scenarios. Our code, data, and demos are available on the [project page](https://knightyxp.github.io/VideoGrain_project_page/).

Accurately modeling chemical reactions using Artificial Intelligence (AI) can accelerate discovery and development, especially in fields like drug design and material science. Although AI has made remarkable advancements in single molecule recognition, such as predicting molecular properties, the study of interactions between molecules, particularly chemical reactions, has been relatively overlooked. In this paper, we introduce Reaction Graph (RG), a unified graph representation that encapsulates the 3D molecular structures within chemical reactions. RG integrates the molecular graphs of reactants and products into a cohesive framework, effectively capturing the interatomic relationships pertinent to the reaction process. Additionally, it incorporates the 3D structure information of molecules in a simple yet effective manner. We conduct experiments on a range of tasks, including chemical reaction classification, condition prediction, and yield prediction. RG achieves the highest accuracy across six datasets, demonstrating its effectiveness. The code is available at https://github.com/Shadow-Dream/Reaction-Graph.

Text-to-3D generation automates 3D content creation from textual descriptions, which offers transformative potential across various fields. However, existing methods often struggle to align generated content with human preferences, limiting their applicability and flexibility. To address these limitations, in this paper, we propose DreamDPO, an optimization-based framework that integrates human preferences into the 3D generation process, through direct preference optimization. Practically, DreamDPO first constructs pairwise examples, then validates their alignment with human preferences using reward or large multimodal models, and lastly optimizes the 3D representation with a preference-driven loss function. By leveraging relative preferences, DreamDPO reduces reliance on precise quality evaluations while enabling fine-grained controllability through preference-guided optimization. Experiments demonstrate that DreamDPO achieves state-of-the-art results, and provides higher-quality and more controllable 3D content compared to existing methods. The code and models will be open-sourced.

Shadow removal aims to restore the image content in shadowed regions. While deep learning-based methods have shown promising results, they still face key challenges: 1) uncontrolled removal of all shadows, or 2) controllable removal but heavily relies on precise shadow region masks. To address these issues, we introduce a novel paradigm: prompt-aware controllable shadow removal. Unlike existing approaches, our paradigm allows for targeted shadow removal from specific subjects based on user prompts (e.g., dots, lines, or subject masks). This approach eliminates the need for shadow annotations and offers flexible, user-controlled shadow removal. Specifically, we propose an end-to-end learnable model, the Prompt-Aware Controllable Shadow Removal Network (PACSRNet). PACSRNet consists of two key modules: a prompt-aware module that generates shadow masks for the specified subject based on the user prompt, and a shadow removal module that uses the shadow prior from the first module to restore the content in the shadowed areas. Additionally, we enhance the shadow removal module by incorporating feature information from the prompt-aware module through a linear operation, providing prompt-guided support for shadow removal. Recognizing that existing shadow removal datasets lack diverse user prompts, we contribute a new dataset specifically designed for prompt-based controllable shadow removal. Extensive experimental results demonstrate the effectiveness and superiority of PACSRNet.

Video inpainting aims to fill the missing regions in video with spatial-temporally coherent contents. Existing methods usually treat the missing contents as a whole and adopt a hybrid objective containing a reconstruction loss and an adversarial loss to train the model. However, these two kinds of loss focus on contents at different frequencies, simply combining them may cause inter-frequency conflicts, leading the trained model to generate compromised results. Inspired by the common corrupted painting restoration process of “drawing a draft first and then revising the details later”, this paper proposes a Drafting-and-Revision Completion Network (DRCN) for video inpainting. Specifically, we first design a Drafting Network that utilizes the temporal information to complete the low-frequency semantic structure at low resolution. Then, a Revision Network is developed to hallucinate high-frequency details at high resolution by using the output of Drafting Network. In this way, adversarial loss and reconstruction loss can be applied to high-frequency and low-frequency respectively, effectively mitigating inter-frequency conflicts. Furthermore, Revision Network can be stacked in a pyramid manner to generate higher resolution details, which provide a feasible solution for high-resolution video inpainting. Experiments show that DRCN achieves improvements of 7.43% and 12.64% in E_warp and LPIPS, and can handle higher resolution videos on limited GPU memory.

Micro-Actions (MAs) are an important form of non-verbal communication in social interactions, with potential applications in human emotional analysis. However, existing methods in Micro-Action Recognition often overlook the inherent subtle changes in MAs, which limits the accuracy of distinguishing MAs with subtle changes. To address this issue, we present a novel Motion-guided Modulation Network (MMN) that implicitly captures and modulates subtle motion cues to enhance spatial-temporal representation learning. Specifically, we introduce a Motion-guided Skeletal Modulation module (MSM) to inject motion cues at the skeletal level, acting as a control signal to guide spatial representation modeling. In parallel, we design a Motion-guided Temporal Modulation module (MTM) to incorporate motion information at the frame level, facilitating the modeling of holistic motion patterns in micro-actions. Finally, we propose a motion consistency learning strategy to aggregate the motion cues from multi-scale features for micro-action classification. Experimental results on the Micro-Action 52 and iMiGUE datasets demonstrate that MMN achieves state-of-the-art performance in skeleton-based micro-action recognition, underscoring the importance of explicitly modeling subtle motion cues. The code will be available at https://github.com/momiji-bit/MMN

Reconstructing transparent surfaces is essential for tasks such as robotic manipulation in labs, yet it poses a significant challenge for 3D reconstruction techniques like 3D Gaussian Splatting (3DGS). These methods often encounter a transparency-depth dilemma, where the pursuit of photorealistic rendering through standard α-blending undermines geometric precision, resulting in considerable depth estimation errors for transparent materials. To address this issue, we introduce Transparent Surface Gaussian Splatting (TSGS), a new framework that separates geometry learning from appearance refinement. In the geometry learning stage, TSGS focuses on geometry by using specular-suppressed inputs to accurately represent surfaces. In the second stage, TSGS improves visual fidelity through anisotropic specular modeling, crucially maintaining the established opacity to ensure geometric accuracy. To enhance depth inference, TSGS employs a first-surface depth extraction method. This technique uses a sliding window over α-blending weights to pinpoint the most likely surface location and calculates a robust weighted average depth. To evaluate the transparent surface reconstruction task under realistic conditions, we collect a TransLab dataset that includes complex transparent laboratory glassware. Extensive experiments on TransLab show that TSGS achieves accurate geometric reconstruction and realistic rendering of transparent objects simultaneously within the efficient 3DGS framework. Specifically, TSGS significantly surpasses current leading methods, achieving a 37.3% reduction in chamfer distance and an 8.0% improvement in F1 score compared to the top baseline. Additionally, TSGS maintains high-quality novel view synthesis, evidenced by a 0.41dB gain in PSNR, demonstrating that TSGS overcomes the transparency-depth dilemma. The code and dataset are available at https://longxiang-ai.github.io/TSGS/.

Protein research is crucial in various scientific disciplines, but understanding their intricate structure-function relationships remains challenging. Recent advancements in Large Language Models (LLMs) have significantly improved the comprehension of task-specific knowledge, suggesting the potential for specialized ChatGPT-like systems in protein research to aid fundamental investigations. In this work, we introduce ProtChatGPT, which aims to learn and understand protein structures using natural language. ProtChatGPT enables users to upload proteins, ask questions, and engage in interactive conversations to produce comprehensive answers. The system comprises multi-level protein encoding, protein-language alignment, and instruction tuning of LLMs. A protein first undergoes multiple protein encoders and PLP-former to produce multi-level hybrid protein embeddings, which are then aligned through a Protein Context Gating (PCG) module with contrastive learning, and projected by an adapter to conform with the LLM. The LLM finally combines user questions with projected protein embeddings to generate informative answers. Experiments show that ProtChatGPT can produce promising responses to proteins and the corresponding user questions. We hope that ProtChatGPT could form the basis for further exploration and application in protein research. Code and our pre-trained model will be publicly available.

Visual reinforcement learning has exhibited efficacy in solving control tasks characterized by high-dimensional observations. However, a central challenge persists in deriving dependable and generalizable representations from vision-based observations. Inspired by the human thought process, when the visual representation extracted from the observation can predict the future and trace history, the representation is reliable and accurate in comprehending the environmental state. Based on this concept, we introduce a B idirectional T ransition (BT) framework for representation learning. This framework employs the bidirectional prediction of both forward and backward environmental transitions as auxiliary tasks to extract reliable representations. Additionally, we introduce an inverse dynamic model to predict the actions causing environmental state transitions, thereby learning the task relevance of state representations. Our method demonstrates competitive generalization performance and sample efficiency in two settings in the DeepMind Control suite. Moreover, we utilize the robotic manipulation simulator, autonomous driving simulator CARLA, and visual navigation simulator Habitat to demonstrate the wide applicability of our method. The results indicate that BT offers more stable and reliable representations and exhibits robust generalization performance for visual reinforcement learning tasks.

We introduce the multimodal task of Protein Captioning , which is an easy-to-understand and flexible way for protein analysis. Compared to specific protein recognition or classification tasks, such as enzyme reaction classification and gene ontology term prediction, protein captioning provides comprehensive textural descriptions for proteins, thus playing a key role in bridging the gap between protein sequences and natural languages. To address the problem, we propose a simple yet effective method, Protein-to-Text Generative Pre-Trained Transformer (P2T-GPT), to fuse multimodal embeddings and translate the chain of amino acid residues in a protein to a sequence of natural language words, i.e., text. For the evaluation of protein captioning, we collect the ProteinCap dataset that contains 94,454 protein-text pairs. Experiments on ProteinCap demonstrate the effectiveness of the proposed P2T-GPT on protein captioning. For example, our method obtains improvements of 8.74, 10.03, and 11.05 in the BERTScore compared to the baseline model on ProteinCap- \(\alpha,\beta,\gamma\) , respectively. As minor contributions, first, P2T-GPT provides a way to connect protein science and Large Language Models (LLMs). By appending ChatGPT, our method can interact in a conversational way to answer questions given a protein. Second, we show that protein captioning can be treated as a pre-trained task that can benefit a range of downstream tasks, to a certain extent.

To achieve content-consistent results in text-conditioned image editing, existing methods typically employ a reconstruction branch to capture the source image details via diffusion inversion and a generation branch to synthesize the target image based on the given textual prompt and the masked source image details. However, accurately segmenting source details is challenging with the current fixed-threshold mask strategy. Additionally, the inadequacies in the inversion process can lead to insufficient retention of source details. In this article, we propose a method called SAMControl (Soft Attention Mask) to adaptively control the pose and object details for image editing. SAMControl dynamically learns flexible attention masks for different images at various diffusion steps. Furthermore, in the reconstruction branch, we utilize a direct inversion technique to ensure the fidelity of source details within SAM. Extensive qualitative and quantitative results demonstrate the effectiveness of the proposed method.

Multimodal Sarcasm Understanding (MSU) has a wide range of applications in the news field such as public opinion analysis and forgery detection. However, existing MSU benchmarks and approaches usually focus on sentence-level MSU. In document-level news, sarcasm clues are sparse or small and are often concealed in long text. Moreover, compared to sentence-level comments like tweets, which mainly focus on only a few trends or hot topics (e.g., sports events), content in the news is considerably diverse. Models created for sentence-level MSU may fail to capture sarcasm clues in document-level news. To fill this gap, we present a comprehensive benchmark for Document-level Multimodal Sarcasm Understanding (DocMSU). Our dataset contains 102,588 pieces of news with text-image pairs, covering 9 diverse topics such as health, business, etc. The proposed large-scale and diverse DocMSU significantly facilitates the research of document-level MSU in real-world scenarios. To take on the new challenges posed by DocMSU, we introduce a fine-grained sarcasm comprehension method to properly align the pixel-level image features with word-level textual features in documents. Experiments demonstrate the effectiveness of our method, showing that it can serve as a baseline approach to the challenging DocMSU.

Hands are the main medium when people interact with the world. Generating proper 3D motion for hand-object interaction is vital for applications such as virtual reality and robotics. Although grasp tracking or object manipulation synthesis can produce coarse hand motion, this kind of motion is inevitably noisy and full of jitter. To address this problem, we propose a data-driven method for coarse motion refinement. First, we design a hand-centric representation to describe the dynamic spatial-temporal relation between hands and objects. Compared to the object-centric representation, our hand-centric representation is straightforward and does not require an ambiguous projection process that converts object-based prediction into hand motion. Second, to capture the dynamic clues of hand-object interaction, we propose a new architecture that models the spatial and temporal structure in a hierarchical manner. Extensive experiments demonstrate that our method outperforms previous methods by a noticeable margin.

Video anomaly understanding (VAU) aims to automat-ically comprehend unusual occurrences in videos, thereby enabling various applications such as traffic surveillance and industrial manufacturing. While existing VAU benchmarks primarily concentrate on anomaly detection and localization, our focus is on more practicality, prompting us to raise the following crucial questions: “what anomaly occurred?”,”why did it happen?”, and “how severe is this abnormal event?”. In pursuit of these answers, we present a comprehensive benchmark for Causation Understanding of Video Anomaly (CUVA). Specifically, each instance of the proposed benchmark involves three sets of human annotations to indicate the”what”, “why” and “how” of an anomaly, including 1) anomaly type, start and end times, and event descriptions, 2) natural language explanations for the cause of an anomaly, and 3) free text reflecting the effect of the abnormality. In addition, we also introduce MMEval, a novel evaluation metric designed to better align with human preferences for CUVA, facilitating the measurement of existing LLMs in comprehending the underlying cause and corresponding effect of video anoma-lies. Finally, we propose a novel prompt-based method that can serve as a baseline approach for the challenging CUVA. We conduct extensive experiments to show the superiority of our evaluation metric and the prompt-based approach. Our code and dataset are available at https://github.com/fesvhtr/CUVA.

Protein representation learning is a challenging task that aims to capture the structure and function of proteins from their amino acid sequences. Previous methods largely ignored the fact that not all amino acids are equally important for protein folding and activity. In this article, we propose a neural clustering framework that can automatically discover the critical components of a protein by considering both its primary and tertiary structure information. Our framework treats a protein as a graph, where each node represents an amino acid and each edge represents a spatial or sequential connection between amino acids. We then apply an iterative clustering strategy to group the nodes into clusters based on their 1D and 3D positions and assign scores to each cluster. We select the highest-scoring clusters and use their medoid nodes for the next iteration of clustering, until we obtain a hierarchical and informative representation of the protein. We evaluate on four protein-related tasks: protein fold classification, enzyme reaction classification, gene ontology term prediction, and enzyme commission number prediction. Experimental results demonstrate that our method achieves state-of-the-art performance.

Previous efforts using frozen Large Language Models (LLMs) for visual understanding, via image captioning or image-text retrieval tasks, face challenges when dealing with complex multimodal scenarios. In order to enhance the capabilities of Multimodal Large Language Models (MLLM) in comprehending the context of vision and language, we introduce Multimodal Composition Learning (MCL) for the purpose of mapping or aligning the vision and language input. In particular, we introduce two tasks: Multimodal-Context Captioning (MC-Cap) and Multimodal-Context Retrieval (MC-Ret) to guide a frozen LLM in comprehending the vision and language context. These specialized tasks are crafted to improve the LLM’s capacity for efficient processing and utilization of multimodal inputs, thereby enhancing its proficiency in generating more accurate text or visual representations. Extensive experiments on both retrieval tasks (i.e., zero-shot composed image retrieval, visual storytelling image retrieval and visual dialog image retrieval) and text generation tasks (i.e., visual question answering) demonstrate the effectiveness of the proposed method. The code is available at: https://github.com/dhg-wei/MCL.

Due to the limitation of collection device and unstable scanning process, point cloud data is usually noisy. This noise deforms the underlying structures of point clouds and inevitably affects downstream tasks such as rendering, reconstruction and classification. In this paper, we propose a Cross-stage Cross-coder Adaptive Edge Graph Convolution Network (C2AENet) to denoise point clouds. Our network uses multiple stages to progressively and iteratively denoise points. To improve the effectiveness, we add connections between two stages and between the encoder and decoder, leading to the cross-stage cross-coder architecture. Additionally, existing graph-based point cloud learning methods tend to capture the local structure. They typically construct a semantic graph based on semantic distance, which may ignore Euclidean neighbors and lead to insufficient geometry perception. Therefore, we introduce a geometric graph and adaptively calculate edge attention based on the local and global structural information of the points. This results in a novel graph convolution module that allows the network to capture richer contextual information and focus on more important parts. Extensive experiments demonstrate that the proposed method is competitive compared with other state-of-the-art methods. The code is available at: https://github.com/chenwuwq/C2AENet.

Recent advancements in image understanding have benefited from the extensive use of web image-text pairs. However, video understanding remains a challenge despite the availability of substantial web video-text data. This difficulty primarily arises from the inherent complexity of videos and the inefficient language supervision in recent web-collected video-text datasets. In this paper, we introduce Text-Only Pre-Alignment (TOPA), a novel approach to extend large language models (LLMs) for video understanding, without the need for pre-training on real video data. Specifically, we first employ an advanced LLM to automatically generate Textual Videos comprising continuous textual frames, along with corresponding annotations to simulate real video-text data. Then, these annotated textual videos are used to pre-align a language-only LLM with the video modality. To bridge the gap between textual and real videos, we employ the CLIP model as the feature extractor to align image and text modalities. During text-only pre-alignment, the continuous textual frames, encoded as a sequence of CLIP text features, are analogous to continuous CLIP image features, thus aligning the LLM with real video representation. Extensive experiments, including zero-shot evaluation and finetuning on various video understanding tasks, demonstrate that TOPA is an effective and efficient framework for aligning video content with LLMs. In particular, without training on any video data, the TOPA-Llama2-13B model achieves a Top-1 accuracy of 51.0% on the challenging long-form video understanding benchmark, Egoschema. This performance surpasses previous video-text pre-training approaches and proves competitive with recent GPT-3.5 based video agents.

The main challenge in video question answering (VideoQA) is to capture and understand the complex spatial and temporal relations between objects based on given questions. Existing graph-based methods for VideoQA usually ignore keywords in questions and employ a simple graph to aggregate features without considering relative relations between objects, which may lead to inferior performance. In this paper, we propose a Keyword-aware Relative Spatio-Temporal (KRST) graph network for VideoQA. First, to make question features aware of keywords, we employ an attention mechanism to assign high weights to keywords during question encoding. The keyword-aware question features are then used to guide video graph construction. Second, because relations are relative, we integrate the relative relation modeling to better capture the spatio-temporal dynamics among object nodes. Moreover, we disentangle the spatio-temporal reasoning into an object-level spatial graph and a frame-level temporal graph, which reduces the impact of spatial and temporal relation reasoning on each other. Extensive experiments on the TGIF-QA, MSVD-QA and MSRVTT-QA datasets demonstrate the superiority of our KRST over multiple state-of-the-art methods.

Learning discriminative and robust representations is important for facial expression recognition (FER) due to subtly different emotional faces and their subjective annotations. Previous works usually address one representation solely because these two goals seem to be contradictory for optimization. Their performances inevitably suffer from challenges from the other representation. In this article, by considering this problem from two novel perspectives, we demonstrate that discriminative and robust representations can be learned in a unified approach, i.e., DR-FER, and mutually benefit each other. Moreover, we make it with the supervision from only original annotations. Specifically, to learn discriminative representations, we propose performing masked image modeling (MIM) as an auxiliary task to force our network to discover expression-related facial areas. This is the first attempt to employ MIM to explore discriminative patterns in a self-supervised manner. To extract robust representations, we present a category-aware self-paced learning schedule to mine high-quality annotated (easy) expressions and incorrectly annotated (hard) counterparts. We further introduce a retrieval similarity-based relabeling strategy to correct hard expression annotations, exploiting them more effectively. By enhancing the discrimination ability of the FER classifier as a bridge, these two learning goals significantly strengthen each other. Extensive experiments on several popular benchmarks demonstrate the superior performance of our DR-FER. Moreover, thorough visualizations and extra experiments on manually annotation-corrupted datasets show that our approach successfully accomplishes learning both discriminative and robust representations simultaneously.

Given an object of interest, visual navigation aims to reach the object’s location based on a sequence of partial observations. To this end, an agent needs to (1) acquire specific knowledge about the relations of object categories in the world during training and (2) locate the target object based on the pre-learned object category relations and its trajectory in the current unseen environment. In this article, we propose a Category Relation Graph (CRG) to learn the knowledge of object category layout relations and a Temporal-Spatial-Region attention (TSR) architecture to perceive the long-term spatial-temporal dependencies of objects, aiding navigation. We establish CRG to learn prior knowledge of object layout and deduce the positions of specific objects. Subsequently, we propose the TSR architecture to capture relationships among objects in temporal, spatial, and regions within observation trajectories. Specifically, we implement a Temporal attention module (T) to model the temporal structure of the observation sequence, implicitly encoding historical moving or trajectory information. Then, a Spatial attention module (S) uncovers the spatial context of the current observation objects based on CRG and past observations. Last, a Region attention module (R) shifts the attention to the target-relevant region. Leveraging the visual representation extracted by our method, the agent accurately perceives the environment and easily learns a superior navigation policy. Experiments on AI2-THOR demonstrate that our CRG-TSR method significantly outperforms existing methods in both effectiveness and efficiency. The supplementary material includes the code and will be publicly available.

Effectively preserving and encoding structure features from objects in irregular and sparse LiDAR points is a crucial challenge to 3D object detection on the point cloud. Recently, Transformer has demonstrated promising performance on many 2D and even 3D vision tasks. Compared with the fixed and rigid convolution kernels, the self-attention mechanism in Transformer can adaptively exclude the unrelated or noisy points and is thus suitable for preserving the local spatial structure in the irregular LiDAR point cloud. However, Transformer only performs a simple sum on the point features, based on the self-attention mechanism, and all the points share the same transformation for value. A such isotropic operation cannot capture the direction-distance-oriented local structure, which is essential for 3D object detection. In this work, we propose a Structure-Embedding transFormer (SEFormer), which can not only preserve the local structure as a traditional Transformer but also have the ability to encode the local structure. Compared to the self-attention mechanism in traditional Transformer, SEFormer learns different feature transformations for value points based on the relative directions and distances to the query point. Then we propose a SEFormer-based network for high-performance 3D object detection. Extensive experiments show that the proposed architecture can achieve SOTA results on the Waymo Open Dataset, one of the most significant 3D detection benchmarks for autonomous driving. Specifically, SEFormer achieves 79.02% mAP, which is 1.2% higher than existing works. https://github.com/tdzdog/SEFormer.

Automatically localizing a position based on a few natural language instructions is essential for future robots to communicate and collaborate with humans. To approach this goal, we focus on a text-to-point-cloud cross-modal localization problem. Given a textual query, it aims to identify the described location from city-scale point clouds. The task involves two challenges. 1) In city-scale point clouds, similar ambient instances may exist in several locations. Searching each location in a huge point cloud with only instances as guidance may lead to less discriminative signals and incorrect results. 2) In textual descriptions, the hints are provided separately. In this case, the relations among those hints are not explicitly described, leaving the difficulties of learning relations to the agent itself. To alleviate the two challenges, we propose a unified Relation-Enhanced Transformer (RET) to improve representation discriminability for both point cloud and nature language queries. The core of the proposed RET is a novel Relation-enhanced Self-Attention (RSA) mechanism, which explicitly encodes instance (hint)-wise relations for the two modalities. Moreover, we propose a fine-grained cross-modal matching method to further refine the location predictions in a subsequent instance-hint matching stage. Experimental results on the KITTI360Pose dataset demonstrate that our approach surpasses the previous state-of-the-art method by large margins.

Deep neural networks on regular 1D lists (e.g., natural languages) and irregular 3D sets (e.g., point clouds) have made tremendous achievements. The key to natural language processing is to model words and their regular order dependency in texts. For point cloud understanding, the challenge is to understand the geometry via irregular point coordinates, in which point-feeding orders do not matter. However, there are a few kinds of data that exhibit both regular 1 D list and irregular 3D set structures, such as proteins and non-coding RNAs. In this paper, we refer to them as 3D point lists and propose a Transformer-style PointListNet to model them. First, PointListNet employs non-parametric distance-based attention because we find sometimes it is the distance, instead of the feature or type, that mainly determines how much two points, e.g., amino acids, are correlated in the micro world. Second, different from the vanilla Transformer that directly performs a simple linear transformation on inputs to generate values and does not explicitly model relative relations, our PointListNet integrates the 1D order and 3D Euclidean displacements into values. We conduct experiments on protein fold classification and enzyme reaction classification. Experimental results show the effectiveness of the proposed PointListNet.

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

Recently, the community has made tremendous progress in developing effective methods for point cloud video understanding that learn from massive amounts of labeled data. However, annotating point cloud videos is usually notoriously expensive. Moreover, training via one or only a few traditional tasks (e.g., classification) may be insufficient to learn subtle details of the spatio-temporal structure existing in point cloud videos. In this paper, we propose a Masked Spatio-Temporal Structure Prediction (MaST-Pre) method to capture the structure of point cloud videos without human annotations. MaST-Pre is based on spatio-temporal point-tube masking and consists of two self-supervised learning tasks. First, by reconstructing masked point tubes, our method is able to capture the appearance information of point cloud videos. Second, to learn motion, we propose a temporal cardinality difference prediction task that estimates the change in the number of points within a point tube. In this way, MaST-Pre is forced to model the spatial and temporal structure in point cloud videos. Extensive experiments on MSRAction-3D, NTU-RGBD, NvGesture, and SHREC’17 demonstrate the effectiveness of the proposed method. The code is available at https://github.com/JohnsonSign/MaST-Pre.

We propose a unified point cloud video self-supervised learning framework for object-centric and scene-centric data. Previous methods commonly conduct representation learning at the clip or frame level and cannot well capture fine-grained semantics. Instead of contrasting the representations of clips or frames, in this paper, we propose a unified self-supervised framework by conducting contrastive learning at the point level. Moreover, we introduce a new pretext task by achieving semantic alignment of superpoints, which further facilitates the representations to capture semantic cues at multiple scales. In addition, due to the high redundancy in the temporal dimension of dynamic point clouds, directly conducting contrastive learning at the point level usually leads to massive undesired negatives and insufficient modeling of positive representations. To remedy this, we propose a selection strategy to retain proper negatives and make use of high-similarity samples from other instances as positive supplements. Extensive experiments show that our method outperforms supervised counterparts on a wide range of downstream tasks and demonstrates the superior transferability of the learned representations.

Due to the inherent unorderliness and irregularity of point cloud, points emerge inconsistently across different frames in a point cloud video. To capture the dynamics in point cloud videos, tracking points and limiting temporal modeling range are usually employed to preserve spatio-temporal structure. However, as points may flow in and out across frames, computing accurate point trajectories is extremely difficult, especially for long videos. Moreover, when points move fast, even in a small temporal window, points may still escape from a region. Besides, using the same temporal range for different motions may not accurately capture the temporal structure. In this paper, we propose a Point Spatio-Temporal Transformer (PST-Transformer). To preserve the spatio-temporal structure, PST-Transformer adaptively searches related or similar points across the entire video by performing self-attention on point features. Moreover, our PST-Transformer is equipped with an ability to encode spatio-temporal structure. Because point coordinates are irregular and unordered but point timestamps exhibit regularities and order, the spatio-temporal encoding is decoupled to reduce the impact of the spatial irregularity on the temporal modeling. By properly preserving and encoding spatio-temporal structure, our PST-Transformer effectively models point cloud videos and shows superior performance on 3D action recognition and 4D semantic segmentation.

In this paper, we propose an unsupervised domain adaptation method for deep point cloud representation learning. To model the internal structures in target point clouds, we first propose to learn the global representations of unla-beled data by scaling up or down point clouds and then predicting the scales. Second, to capture the local structure in a self-supervised manner, we propose to project a 3D local area onto a 2D plane and then learn to reconstruct the squeezed region. Moreover, to effectively transfer the knowledge from source domain, we propose to vote pseudo labels for target samples based on the labels of their nearest source neighbors in the shared feature space. To avoid the noise caused by incorrect pseudo labels, we only select re-liable target samples, whose voting consistencies are high enough, for enhancing adaptation. The voting method is able to adaptively select more and more target samples during training, which in return facilitates adaptation because the amount of labeled target data increases. Experiments on PointDA (ModelNet-10, ShapeNet-10 and ScanNet-10) and Sim-to-Real (ModelNet-11, ScanObjectNN-11, ShapeNet-9 and ScanObjectNN-9) demonstrate the effectiveness of our method.

In point cloud videos, point coordinates are irregular and unordered but point timestamps exhibit regularities and order. Grid-based networks for conventional video processing cannot be directly used to model raw point cloud videos. Therefore, in this work, we propose a point-based network that directly handles raw point cloud videos. First, to preserve the spatio-temporal local structure of point cloud videos, we design a point tube covering a local range along spatial and temporal dimensions. By progressively subsampling frames and points and enlarging the spatial radius as the point features are fed into higher-level layers, the point tube can capture video structure in a spatio-temporally hierarchical manner. Second, to reduce the impact of the spatial irregularity on temporal modeling, we decompose space and time when extracting point tube representations. Specifically, a spatial operation is employed to encode the local structure of each spatial region in a tube and a temporal operation is used to encode the dynamics of the spatial regions along the tube. Empirically, the proposed network shows strong performance on 3D action recognition, 4D semantic segmentation and scene flow estimation. Theoretically, we analyse the necessity to decompose space and time in point cloud video modeling and why the network outperforms existing methods.

Hand-object interaction plays a very important role when humans manipulate objects. While existing methods focus on improving hand-object recognition with fully automatic methods, human intention has been largely neglected in the recognition process, thus leading to undesirable interaction descriptions. To better interpret human-object interaction that is aligned to human intention, we argue that a reference specifying human intention should be taken into account. Thus, we propose a new approach to represent interactions while reflecting human purpose with three key factors,i.e., hand, object and reference. Specifically, we design a pattern ofhand-object, object-reference, hand, object, reference> (HOR) to recognize intention based atomic hand-object interactions. This pattern aims to model interactions with the states of hand, object, reference and their relationships. Furthermore, we design a simple yet effective Spatially Part-based (3+1)D convolutional neural network, namely SP(3+1)D, which leverages 3D and 1D convolutions to model visual dynamics and object position changes based on our HOR, respectively. With the help of our SP(3+1)D network, the recognition results are able to indicate human purposes accurately. To evaluate the proposed method, we annotate a Something-1.3k dataset, which contains 10 atomic hand-object interactions and about 130 videos for each interaction. Experimental results on Something-1.3k demonstrate the effectiveness of our SP(3+1)D network.

The superiority of deeply learned representations relies on large-scale labeled datasets. However, annotating data are usually expensive or even infeasible in some scenarios. To address this problem, we propose an unsupervised method to leverage instance discrimination and similarity for deep visual representation learning. The method is based on an observation that convolutional neural networks (CNNs) can learn a meaningful visual representation with instancewise classification, in which each instance is treated as an individual class. By this instancewise discriminative learning, instances can reasonably distribute in the representation space, which reveals their similarities. In order to further improve visual representations, we propose a dual-level progressive similar instance selection (DPSIS) method to build a bridge from instance to class by selecting similar instances (neighbors) for each instance (anchor) and treating the anchor and its neighbors as the same class. To be specific, DPSIS adaptively selects two levels of neighbors, that is: 1) an "absolutely similar level" and 2) a "relatively similar level." Instances in the absolutely similar level are used as hard labels, while instances in the relatively similar level are used as soft labels. Moreover, during training, DPSIS is able to progressively select more neighbors without human supervision. At the beginning of training, because CNNs are weak, most instances are distributed relatively randomly in the representation space and only a few easy-to-recognize instances are selected as neighbors. As CNN models become stronger, the semantic meaning of each instance grows clearer. Those instances originally distributed in a relatively random manner gradually move to meaningful positions. This consequently facilitates CNN training since the number of reliable samples increases. Experiments on seven benchmarks, including three small-scale and two large-scale coarse-grained image classification datasets, and two fine-grained categorization datasets, demonstrate the effectiveness of our DPSIS. Our codes have been released at https://github.com/hehefan/DPSIS.

Neighboring frames are more correlated compared to frames from further temporal distances. In this paper, we aim to explore the temporal correlations among neighboring frames and exploit cross-layer multi-scale features for action recognition. First, we present a Temporal Cross-Layer Correlation (TCLC) framework for temporal correlation learning. The unified framework uncovers both local and global structures from video data, enabling a better exploration of temporal context and assisting cross-layer spatio-temporal feature learning. Second, we propose a novel cross-layer attention and a center-guided attention mechanism to integrate features with contextual knowledge from multiple scales. Our method is a two-stage process for effective cross-layer feature learning. The first stage incorporates the cross-layer attention module to decide the importance weight of the convolutional layers. The second stage leverages the center-guided attention mechanism to aggregate local features from each layer for the generation of a final video representation. We leverage global centers to extract shared semantic knowledge among videos. We evaluate TCLC on three action recognition datasets, i.e., UCF-101, HMDB-51 and Kinetics. Our experimental results demonstrate the superiority of our proposed temporal correlation mining method.

Point cloud videos exhibit irregularities and lack of order along the spatial dimension where points emerge inconsistently across different frames. To capture the dynamics in point cloud videos, point tracking is usually employed. However, as points may flow in and out across frames, computing accurate point trajectories is extremely difficult. Moreover, tracking usually relies on point colors and thus may fail to handle colorless point clouds. In this paper, to avoid point tracking, we propose a novel Point 4D Transformer (P4Transformer) network to model raw point cloud videos. Specifically, P4Transformer consists of (i) a point 4D convolution to embed the spatio-temporal local structures presented in a point cloud video and (ii) a transformer to capture the appearance and motion information across the entire video by performing self-attention on the embedded local features. In this fashion, related or similar local areas are merged with attention weight rather than by explicit tracking. Extensive experiments, including 3D action recognition and 4D semantic segmentation, on four benchmarks demonstrate the effectiveness of our P4Transformer for point cloud video modeling.

Motion, according to its definition in physics, is the change in position with respect to time, regardless of the specific moving object and background. In this paper, we aim to learn appearance-independent motion representation in an unsupervised manner. The main idea is to separate motion from videos while leaving objects and background as content. Specifically, we design an encoder-decoder model which consists of a content encoder, a motion encoder and a video generator. To train the model, we leverage a one-step cycle-consistency in reconstruction within the same video and a two-step cycle-consistency in generation across different videos as self-supervised signals, and use adversarial training to remove the content representation from the motion representation. We demonstrate that the proposed framework can be used for conditional video generation and fine-grained action recognition.

In the few-shot common-localization task, given few support images without bounding box annotations at each episode, the goal is to localize the common object in the query image of unseen categories. The few-shot common-localization task involves common object reasoning from the given images, predicting the spatial locations of the object with different shapes, sizes, and orientations. In this work, we propose a common-centric localization (CCL) network for few-shot common-localization. The motivation of our common-centric localization network is to learn the common object features by dynamic feature relation reasoning via a graph convolutional network with conditional feature aggregation. First, we propose a local common object region generation pipeline to reduce background noises due to feature misalignment. Each support image predicts more accurate object spatial locations by replacing the query with the images in the support set. Second, we introduce a graph convolutional network with dynamic feature transformation to enforce the common object reasoning. To enhance the discriminability during feature matching and enable a better generalization in unseen scenarios, we leverage a conditional feature encoding function to alter visual features according to the input query adaptively. Third, we introduce a common-centric relation structure to model the correlation between the common features and the query image feature. The generated common features guide the query image feature towards a more common object-related representation. We evaluate our common-centric localization network on four datasets, i.e., CL-VOC-07, CL-VOC-12, CL-COCO, CL-VID. We obtain significant improvements compared to state-of-the-art. Our quantitative results confirm the effectiveness of our network.

This paper focuses on the problem of person tube (a sequence of bounding boxes which encloses a person in a video) retrieval using a natural language query. Different from images in person re-identification (re-ID) or person search, besides appearance, person tube contains abundant action and information. We exploit a 2D and a 3D residual networks (ResNets) to extract the appearance and action representation, respectively. To transform tubes and descriptions into a shared latent space where data from the two different modalities can be compared directly, we propose a Multi-Scale Structure Preservation (MSSP) approach. MSSP splits a person tube into several element-tubes on average, whose features are extracted by the two ResNets. Any number of consecutive element-tubes forms a sub-tube. MSSP considers the following constraints for sub-tubes and descriptions in the shared space. 1) Bidirectional ranking. Matching sub-tubes (resp. descriptions) should get ranked higher than incorrect ones for each description (resp. sub-tube). 2) External structure preservation. Sub-tubes (resp. descriptions) from different persons should stay away from each other. 3) Internal structure preservation. Sub-tubes (resp. descriptions) from the same person should be close to each other. Experimental results on person tube retrieval via language description and other two related tasks demonstrate the efficacy of MSSP.

Image captioning, a challenging task where the machine automatically describes an image with natural language, has drawn significant attention in recent years. Despite the remarkable improvements of recent approaches, however, these methods are built upon a large set of training image-sentence pairs. The expensive labor efforts hence limit the captioning model to describe the wider world. In this paper, we present a novel network structure, Cascaded Revision Network, which aims at relieving the problem by equipping the model with out-of-domain knowledge. CRN first tries its best to describe an image using the existing vocabulary from in-domain knowledge. Due to the lack of out-of-domain knowledge, the caption may be inaccurate or include ambiguous words for the image with unknown (novel) objects. We propose to re-edit the primary captioning sentence by a series of cascaded operations. We introduce a perplexity predictor to find out which words are most likely to be inaccurate given the input image. Thereafter, we utilize external knowledge from a pretrained object detection model and select more accurate words from detection results by the visual matching module. In the last step, we design a semantic matching module to ensure that the novel object is fit in the right position. By this novel cascaded captioning-revising mechanism, CRN can accurately describe images with unseen objects. We validate the proposed method with state-of-the-art performance on the held-out MSCOCO dataset as well as scale to ImageNet, demonstrating the effectiveness of our method.

Due to domain bias, directly deploying a deep person re-identification (re-ID) model trained on one dataset often achieves considerably poor accuracy on another dataset. In this article, we propose an Adaptive Exploration (AE) method to address the domain-shift problem for re-ID in an unsupervised manner. Specifically, in the target domain, the re-ID model is inducted to (1) maximize distances between all person images and (2) minimize distances between similar person images. In the first case, by treating each person image as an individual class, a non-parametric classifier with a feature memory is exploited to encourage person images to move far away from each other. In the second case, according to a similarity threshold, our method adaptively selects neighborhoods for each person image in the feature space. By treating these similar person images as the same class, the non-parametric classifier forces them to stay closer. However, a problem of the adaptive selection is that, when an image has too many neighborhoods, it is more likely to attract other images as its neighborhoods. As a result, a minority of images may select a large number of neighborhoods while a majority of images has only a few neighborhoods. To address this issue, we additionally integrate a balance strategy into the adaptive selection. We evaluate our methods with two protocols. The first one is called “target-only re-ID”, in which only the unlabeled target data is used for training. The second one is called “domain adaptive re-ID”, in which both the source data and the target data are used during training. Experimental results on large-scale re-ID datasets demonstrate the effectiveness of our method. Our code has been released at https://github.com/dyh127/Adaptive-Exploration-for-Unsupervised-Person-Re-Identification.

In Visual Dialog, an agent has to parse temporal context in the dialog history and spatial context in the image to hold a meaningful dialog with humans. For example, to answer “what is the man on her left wearing?” the agent needs to (1) analyze the temporal context in the dialog history to infer who is being referred to as “her,” (2) parse the image to attend “her,” and (3) uncover the spatial context to shift the attention to “her left” and check the apparel of the man. In this article, we use a dialog network to memorize the temporal context and an attention processor to parse the spatial context. Since the question and the image are usually very complex, which makes it difficult for the question to be grounded with a single glimpse, the attention processor attends to the image multiple times to better collect visual information. In the Visual Dialog task, the generative decoder (G) is trained under the word-by-word paradigm, which suffers from the lack of sentence-level training. We propose to reinforce G at the sentence level using the discriminative model (D), which aims to select the right answer from a few candidates, to ameliorate the problem. Experimental results on the VisDial dataset demonstrate the effectiveness of our approach.

Predicting future frames in videos has become a promising direction of research for both computer vision and robot learning communities. The core of this problem involves moving object capture and future motion prediction. While object capture specifies which objects are moving in videos, motion prediction describes their future dynamics. Motivated by this analysis, we propose a Cubic Long Short-Term Memory (CubicLSTM) unit for video prediction. CubicLSTM consists of three branches, i.e., a spatial branch for capturing moving objects, a temporal branch for processing motions, and an output branch for combining the first two branches to generate predicted frames. Stacking multiple CubicLSTM units along the spatial branch and output branch, and then evolving along the temporal branch can form a cubic recurrent neural network (CubicRNN). Experiment shows that CubicRNN produces more accurate video predictions than prior methods on both synthetic and real-world datasets.

Open set domain adaptation aims to diminish the domain shift across domains, with partially shared classes. There exist unknown target samples out of the knowledge of source domain. Compared to the close set setting, how to separate the unknown (unshared) class from the known (shared) ones plays the key role. Whereas, previous methods did not emphasize the semantic structure of the open set data, which may introduce bias into the domain alignment and confuse the classifier around the decision boundary. In this paper, we exploit the semantic structure of open set data from two aspects: 1) Semantic Categorical Alignment, which aims to achieve good separability of target known classes by categorically aligning the centroid of target with the source. 2) Semantic Contrastive Mapping, which aims to push the unknown class away from the decision boundary. Empirically, we demonstrate that our method performs favourably against the state-of-the-art methods on representative benchmarks, e.g. Digits and Office-31 datasets.

We aim to significantly reduce the computational cost for classification of temporally untrimmed videos while retaining similar accuracy. Existing video classification methods sample frames with a predefined frequency over entire video. Differently, we propose an end-to-end deep reinforcement approach which enables an agent to classify videos by watching a very small portion of frames like what we do. We make two main contributions. First, information is not equally distributed in video frames along time. An agent needs to watch more carefully when a clip is informative and skip the frames if they are redundant or irrelevant. The proposed approach enables the agent to adapt sampling rate to video content and skip most of the frames without the loss of information. Second, in order to have a confident decision, the number of frames that should be watched by an agent varies greatly from one video to another. We incorporate an adaptive stop network to measure confidence score and generate timely trigger to stop the agent watching videos, which improves efficiency without loss of accuracy. Our approach reduces the computational cost significantly for the large-scale YouTube-8M dataset, while the accuracy remains the same.

The superiority of deeply learned pedestrian representations has been reported in very recent literature of person re-identification (re-ID). In this article, we consider the more pragmatic issue of learning a deep feature with no or only a few labels. We propose a progressive unsupervised learning (PUL) method to transfer pretrained deep representations to unseen domains. Our method is easy to implement and can be viewed as an effective baseline for unsupervised re-ID feature learning. Specifically, PUL iterates between (1) pedestrian clustering and (2) fine-tuning of the convolutional neural network (CNN) to improve the initialization model trained on the irrelevant labeled dataset. Since the clustering results can be very noisy, we add a selection operation between the clustering and fine-tuning. At the beginning, when the model is weak, CNN is fine-tuned on a small amount of reliable examples that locate near to cluster centroids in the feature space. As the model becomes stronger, in subsequent iterations, more images are being adaptively selected as CNN training samples. Progressively, pedestrian clustering and the CNN model are improved simultaneously until algorithm convergence. This process is naturally formulated as self-paced learning. We then point out promising directions that may lead to further improvement. Extensive experiments on three large-scale re-ID datasets demonstrate that PUL outputs discriminative features that improve the re-ID accuracy. Our code has been released at https://github.com/hehefan/Unsupervised-Person-Re-identification-Clustering-and-Fine-tuning.

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