Sibei Yang

Zhenxiang Lin, Xidong Peng, Peishan Cong et al.

We introduce the task of 3D visual grounding in large-scale dynamic scenes based on natural linguistic descriptions and online captured multi-modal visual data, including 2D images and 3D LiDAR point clouds. We present a novel method, dubbed WildRefer, for this task by fully utilizing the rich appearance information in images, the position and geometric clues in point cloud as well as the semantic knowledge of language descriptions. Besides, we propose two novel datasets, i.e., STRefer and LifeRefer, which focus on large-scale human-centric daily-life scenarios accompanied with abundant 3D object and natural language annotations. Our datasets are significant for the research of 3D visual grounding in the wild and has huge potential to boost the development of autonomous driving and service robots. Extensive experiments and ablation studies demonstrate that our method achieves state-of-the-art performance on the proposed benchmarks. The code is provided in https://github.com/4DVLab/WildRefer.

Meng Lou, Shu Zhang, Hong-Yu Zhou et al.

Recent studies have integrated convolutions into transformers to introduce inductive bias and improve generalization performance. However, the static nature of conventional convolution prevents it from dynamically adapting to input variations, resulting in a representation discrepancy between convolution and self-attention as the latter computes attention maps dynamically. Furthermore, when stacking token mixers that consist of convolution and self-attention to form a deep network, the static nature of convolution hinders the fusion of features previously generated by self-attention into convolution kernels. These two limitations result in a sub-optimal representation capacity of the entire network. To find a solution, we propose a lightweight Dual Dynamic Token Mixer (D-Mixer) to simultaneously learn global and local dynamics via computing input-dependent global and local aggregation weights. D-Mixer works by applying an efficient global attention module and an input-dependent depthwise convolution separately on evenly split feature segments, endowing the network with strong inductive bias and an enlarged receptive field. We use D-Mixer as the basic building block to design TransXNet, a novel hybrid CNN-Transformer vision backbone network that delivers compelling performance. In the ImageNet-1K classification, TransXNet-T surpasses Swin-T by 0.3% in top-1 accuracy while requiring less than half of the computational cost. Furthermore, TransXNet-S and TransXNet-B exhibit excellent model scalability, achieving top-1 accuracy of 83.8% and 84.6% respectively, with reasonable computational costs. Additionally, our proposed network architecture demonstrates strong generalization capabilities in various dense prediction tasks, outperforming other state-of-the-art networks while having lower computational costs. Code is publicly available at https://github.com/LMMMEng/TransXNet.

Chaoqi Chen, Yushuang Wu, Qiyuan Dai et al.

Graph Neural Networks (GNNs) have gained momentum in graph representation learning and boosted the state of the art in a variety of areas, such as data mining (\emph{e.g.,} social network analysis and recommender systems), computer vision (\emph{e.g.,} object detection and point cloud learning), and natural language processing (\emph{e.g.,} relation extraction and sequence learning), to name a few. With the emergence of Transformers in natural language processing and computer vision, graph Transformers embed a graph structure into the Transformer architecture to overcome the limitations of local neighborhood aggregation while avoiding strict structural inductive biases. In this paper, we present a comprehensive review of GNNs and graph Transformers in computer vision from a task-oriented perspective. Specifically, we divide their applications in computer vision into five categories according to the modality of input data, \emph{i.e.,} 2D natural images, videos, 3D data, vision + language, and medical images. In each category, we further divide the applications according to a set of vision tasks. Such a task-oriented taxonomy allows us to examine how each task is tackled by different GNN-based approaches and how well these approaches perform. Based on the necessary preliminaries, we provide the definitions and challenges of the tasks, in-depth coverage of the representative approaches, as well as discussions regarding insights, limitations, and future directions.

Hong-Yu Zhou, Chixiang Lu, Chaoqi Chen et al.

Recent advances in self-supervised learning (SSL) in computer vision are primarily comparative, whose goal is to preserve invariant and discriminative semantics in latent representations by comparing siamese image views. However, the preserved high-level semantics do not contain enough local information, which is vital in medical image analysis (e.g., image-based diagnosis and tumor segmentation). To mitigate the locality problem of comparative SSL, we propose to incorporate the task of pixel restoration for explicitly encoding more pixel-level information into high-level semantics. We also address the preservation of scale information, a powerful tool in aiding image understanding but has not drawn much attention in SSL. The resulting framework can be formulated as a multi-task optimization problem on the feature pyramid. Specifically, we conduct multi-scale pixel restoration and siamese feature comparison in the pyramid. In addition, we propose non-skip U-Net to build the feature pyramid and develop sub-crop to replace multi-crop in 3D medical imaging. The proposed unified SSL framework (PCRLv2) surpasses its self-supervised counterparts on various tasks, including brain tumor segmentation (BraTS 2018), chest pathology identification (ChestX-ray, CheXpert), pulmonary nodule detection (LUNA), and abdominal organ segmentation (LiTS), sometimes outperforming them by large margins with limited annotations.

Hanzhuo Huang, Yufan Feng, Cheng Shi et al.

Text-to-video is a rapidly growing research area that aims to generate a semantic, identical, and temporal coherence sequence of frames that accurately align with the input text prompt. This study focuses on zero-shot text-to-video generation considering the data- and cost-efficient. To generate a semantic-coherent video, exhibiting a rich portrayal of temporal semantics such as the whole process of flower blooming rather than a set of "moving images", we propose a novel Free-Bloom pipeline that harnesses large language models (LLMs) as the director to generate a semantic-coherence prompt sequence, while pre-trained latent diffusion models (LDMs) as the animator to generate the high fidelity frames. Furthermore, to ensure temporal and identical coherence while maintaining semantic coherence, we propose a series of annotative modifications to adapting LDMs in the reverse process, including joint noise sampling, step-aware attention shift, and dual-path interpolation. Without any video data and training requirements, Free-Bloom generates vivid and high-quality videos, awe-inspiring in generating complex scenes with semantic meaningful frame sequences. In addition, Free-Bloom is naturally compatible with LDMs-based extensions.

Cheng Shi, Yulin Zhang, Bin Yang et al.

Unsupervised 3D instance segmentation aims to segment objects from a 3D point cloud without any annotations. Existing methods face the challenge of either too loose or too tight clustering, leading to under-segmentation or over-segmentation. To address this issue, we propose Part2Object, hierarchical clustering with object guidance. Part2Object employs multi-layer clustering from points to object parts and objects, allowing objects to manifest at any layer. Additionally, it extracts and utilizes 3D objectness priors from temporally consecutive 2D RGB frames to guide the clustering process. Moreover, we propose Hi-Mask3D to support hierarchical 3D object part and instance segmentation. By training Hi-Mask3D on the objects and object parts extracted from Part2Object, we achieve consistent and superior performance compared to state-of-the-art models in various settings, including unsupervised instance segmentation, data-efficient fine-tuning, and cross-dataset generalization. Code is release at https://github.com/ChengShiest/Part2Object

Cheng Shi, Yuchen Zhu, Sibei Yang

Recent advancements in large-scale foundational models have sparked widespread interest in training highly proficient large vision models. A common consensus revolves around the necessity of aggregating extensive, high-quality annotated data. However, given the inherent challenges in annotating dense tasks in computer vision, such as object detection and segmentation, a practical strategy is to combine and leverage all available data for training purposes. In this work, we propose Plain-Det, which offers flexibility to accommodate new datasets, robustness in performance across diverse datasets, training efficiency, and compatibility with various detection architectures. We utilize Def-DETR, with the assistance of Plain-Det, to achieve a mAP of 51.9 on COCO, matching the current state-of-the-art detectors. We conduct extensive experiments on 13 downstream datasets and Plain-Det demonstrates strong generalization capability. Code is release at https://github.com/ChengShiest/Plain-Det

Jiajin Tang, Ge Zheng, Sibei Yang

Referring video object segmentation aims to segment a referent throughout a video sequence according to a natural language expression. It requires aligning the natural language expression with the objects' motions and their dynamic associations at the global video level but segmenting objects at the frame level. To achieve this goal, we propose to simultaneously maintain a global referent token and a sequence of object queries, where the former is responsible for capturing video-level referent according to the language expression, while the latter serves to better locate and segment objects with each frame. Furthermore, to explicitly capture object motions and spatial-temporal cross-modal reasoning over objects, we propose a novel temporal collection-distribution mechanism for interacting between the global referent token and object queries. Specifically, the temporal collection mechanism collects global information for the referent token from object queries to the temporal motions to the language expression. In turn, the temporal distribution first distributes the referent token to the referent sequence across all frames and then performs efficient cross-frame reasoning between the referent sequence and object queries in every frame. Experimental results show that our method outperforms state-of-the-art methods on all benchmarks consistently and significantly.

Yu Wu, Yana Wei, Haozhe Wang et al.

This paper introduces Grounded Image Text Matching with Mismatched Relation (GITM-MR), a novel visual-linguistic joint task that evaluates the relation understanding capabilities of transformer-based pre-trained models. GITM-MR requires a model to first determine if an expression describes an image, then localize referred objects or ground the mismatched parts of the text. We provide a benchmark for evaluating pre-trained models on this task, with a focus on the challenging settings of limited data and out-of-distribution sentence lengths. Our evaluation demonstrates that pre-trained models lack data efficiency and length generalization ability. To address this, we propose the Relation-sensitive Correspondence Reasoning Network (RCRN), which incorporates relation-aware reasoning via bi-directional message propagation guided by language structure. RCRN can be interpreted as a modular program and delivers strong performance in both length generalization and data efficiency.

Zhendong He, Qiyuan Dai, Guanbin Li et al.

Large Vision-Language Models (LVLMs) are rapidly evolving toward true multimodal reasoning, with visual search representing a concrete instantiation of the thinking-with-images paradigm. However, LVLM visual search faces two key challenges: incompatibility among intrinsic capabilities after post-training, and interference in long multi-step reasoning contexts. To address these, we identify two novel insights. First, self-regulation between pre- and post-training LVLMs leverages the intrinsic single-step capabilities of the pre-training model to mitigate capability deterioration and long-context interference. Second, probability-based prophetic sampling, replacing naive prompting, provides a probabilistic interface where the pre-training model acts as a prophet and the post-training model selectively accepts prophetic tokens under its output distribution, preserving coherent multi-step reasoning. Building on these insights, we introduce SeProD, a self-prophetic decoding framework that leverages intrinsic single-step capabilities to enable coherent multi-step reasoning in a training-free, plug-and-play manner. Experiments show that SeProD consistently improves multiple visual-search LVLMs across all 12 splits of 4 visual search benchmarks, as well as across general VQA benchmarks, without added computational overhead, thanks to its parallel prophetic acceptance mechanism.

Ge Zheng, Bin Yang, Jiajin Tang et al.

A long-standing goal of AI systems is to perform complex multimodal reasoning like humans. Recently, large language models (LLMs) have made remarkable strides in such multi-step reasoning on the language modality solely by leveraging the chain of thought (CoT) to mimic human thinking. However, the transfer of these advancements to multimodal contexts introduces heightened challenges, including but not limited to the impractical need for labor-intensive annotation and the limitations in terms of flexibility, generalizability, and explainability. To evoke CoT reasoning in multimodality, this work first conducts an in-depth analysis of these challenges posed by multimodality and presents two key insights: "keeping critical thinking" and "letting everyone do their jobs" in multimodal CoT reasoning. Furthermore, this study proposes a novel DDCoT prompting that maintains a critical attitude through negative-space prompting and incorporates multimodality into reasoning by first dividing the reasoning responsibility of LLMs into reasoning and recognition and then integrating the visual recognition capability of visual models into the joint reasoning process. The rationales generated by DDCoT not only improve the reasoning abilities of both large and small language models in zero-shot prompting and fine-tuning learning, significantly outperforming state-of-the-art methods but also exhibit impressive generalizability and explainability.

Ruifei Zhang, Wei Zhang, Xiao Tan et al.

Recent advancements in language-grounded autonomous driving have been significantly promoted by the sophisticated cognition and reasoning capabilities of large language models (LLMs). However, current LLM-based approaches encounter critical challenges: (1) Failure analysis reveals that frequent collisions and obstructions, stemming from limitations in visual representations, remain primary obstacles to robust driving performance. (2) The substantial parameters of LLMs pose considerable deployment hurdles. To address these limitations, we introduce VLDrive, a novel approach featuring a lightweight MLLM architecture with enhanced vision components. VLDrive achieves compact visual tokens through innovative strategies, including cycle-consistent dynamic visual pruning and memory-enhanced feature aggregation. Furthermore, we propose a distance-decoupled instruction attention mechanism to improve joint visual-linguistic feature learning, particularly for long-range visual tokens. Extensive experiments conducted in the CARLA simulator demonstrate VLDrive`s effectiveness. Notably, VLDrive achieves state-of-the-art driving performance while reducing parameters by 81% (from 7B to 1.3B), yielding substantial driving score improvements of 15.4%, 16.8%, and 7.6% at tiny, short, and long distances, respectively, in closed-loop evaluations. Code is available at https://github.com/ReaFly/VLDrive.

Yu Pan, Andi Zhang, Yi Wang et al.

Diffusion Vision-Language Models (dVLMs), built upon the non-causal foundations of Diffusion Large Language Models (dLLMs), have demonstrated remarkable efficacy in multimodal tasks by departing from the traditional autoregressive generation paradigm. While dVLMs appear inherently robust against conventional jailbreak tactics, which we categorize as Fixed Prefix Optimization (FPO) (e.g., anchoring responses with "Sure, here is"), this perceived resilience is deceptive. Our investigation into the safety landscape of dVLMs reveals a unique refusal pattern: Immediate Refusal and Progressive Refusal. We find that while FPO-based attacks often fail by triggering the latter, the progressive refinement process itself uncovers a novel, latent attack surface. To exploit this vulnerability, we propose Global Probability Optimization (GPO), a general jailbreak paradigm designed specifically for the denoising trajectory of masked diffusion models. Unlike prefix-based methods, GPO manipulates the global generative dynamics to bypass guardrails in diffusion language models. Building on this, we introduce GPO-V, the first visual-modality jailbreak framework tailored for dVLMs. Empirical results demonstrate that GPO-V produces stealthy perturbations with exceptional cross-model transferability, revealing a critical security gap in non-sequential generative architectures. Our findings underscore the critical urgency of addressing safety alignment in dVLMs. These results necessitate an immediate and fundamental re-evaluation of current defense paradigms to mitigate the unique risks of diffusion-based generation. Our code is available at: https://anonymous.4open.science/r/GPO-V-0250.

Cheng Shi, Yizhou Yu, Sibei Yang

Vision Transformers (ViTs), when pre-trained on large-scale data, provide general-purpose representations for diverse downstream tasks. However, artifacts in ViTs are widely observed across different supervision paradigms and downstream tasks. Through systematic analysis of artifacts in ViTs, we find that their fundamental mechanisms have yet to be sufficiently elucidated. In this paper, through systematic analysis, we conclude that these artifacts originate from a lazy aggregation behavior: ViT uses semantically irrelevant background patches as shortcuts to represent global semantics, driven by global attention and Coarse-grained semantic supervision. Our solution selectively integrates patch features into the CLS token, reducing the influence of background-dominated shortcuts and consistently improving performance across 12 benchmarks under label-, text-, and self-supervision. We hope this work offers a new perspective on ViT behavior.

Yuchen Zhu, Cheng Shi, Dingyou Wang et al.

Class-incremental/Continual image segmentation (CIS) aims to train an image segmenter in stages, where the set of available categories differs at each stage. To leverage the built-in objectness of query-based transformers, which mitigates catastrophic forgetting of mask proposals, current methods often decouple mask generation from the continual learning process. This study, however, identifies two key issues with decoupled frameworks: loss of plasticity and heavy reliance on input data order. To address these, we conduct an in-depth investigation of the built-in objectness and find that highly aggregated image features provide a shortcut for queries to generate masks through simple feature alignment. Based on this, we propose SimCIS, a simple yet powerful baseline for CIS. Its core idea is to directly select image features for query assignment, ensuring "perfect alignment" to preserve objectness, while simultaneously allowing queries to select new classes to promote plasticity. To further combat catastrophic forgetting of categories, we introduce cross-stage consistency in selection and an innovative "visual query"-based replay mechanism. Experiments demonstrate that SimCIS consistently outperforms state-of-the-art methods across various segmentation tasks, settings, splits, and input data orders. All models and codes will be made publicly available at https://github.com/SooLab/SimCIS.

Cheng Shi, Sibei Yang

Foundation models, pre-trained on a large amount of data have demonstrated impressive zero-shot capabilities in various downstream tasks. However, in object detection and instance segmentation, two fundamental computer vision tasks heavily reliant on extensive human annotations, foundation models such as SAM and DINO struggle to achieve satisfactory performance. In this study, we reveal that the devil is in the object boundary, \textit{i.e.}, these foundation models fail to discern boundaries between individual objects. For the first time, we probe that CLIP, which has never accessed any instance-level annotations, can provide a highly beneficial and strong instance-level boundary prior in the clustering results of its particular intermediate layer. Following this surprising observation, we propose $\textbf{Zip}$ which $\textbf{Z}$ips up CL$\textbf{ip}$ and SAM in a novel classification-first-then-discovery pipeline, enabling annotation-free, complex-scene-capable, open-vocabulary object detection and instance segmentation. Our Zip significantly boosts SAM's mask AP on COCO dataset by 12.5% and establishes state-of-the-art performance in various settings, including training-free, self-training, and label-efficient finetuning. Furthermore, annotation-free Zip even achieves comparable performance to the best-performing open-vocabulary object detecters using base annotations. Code is released at https://github.com/ChengShiest/Zip-Your-CLIP

Yulin Zhang, Cheng Shi, Sibei Yang

Recent advances in Multimodal Large Language Models have greatly improved visual understanding and reasoning, yet their quadratic attention and offline training protocols make them ill-suited for streaming settings where frames arrive sequentially and future observations are inaccessible. We diagnose a core limitation of current Video-LLMs, namely Time-Agnosticism, in which videos are treated as an unordered bag of evidence rather than a causally ordered sequence, yielding two failures in streams: temporal order ambiguity, in which the model cannot follow or reason over the correct chronological order, and past-current focus blindness where it fails to distinguish present observations from accumulated history. We present WeaveTime, a simple, efficient, and model agnostic framework that first teaches order and then uses order. We introduce a lightweight Temporal Reconstruction objective-our Streaming Order Perception enhancement-that instills order aware representations with minimal finetuning and no specialized streaming data. At inference, a Past-Current Dynamic Focus Cache performs uncertainty triggered, coarse-to-fine retrieval, expanding history only when needed. Plugged into exsiting Video-LLM without architectural changes, WeaveTime delivers consistent gains on representative streaming benchmarks, improving accuracy while reducing latency. These results establish WeaveTime as a practical path toward time aware stream Video-LLMs under strict online, time causal constraints. Code and weights will be made publicly available. Project Page: https://zhangyl4.github.io/publications/weavetime/

Yifan Wang, Yifei Liu, Yingdong Shi et al.

Vision Transformer models exhibit immense power yet remain opaque to human understanding, posing challenges and risks for practical applications. While prior research has attempted to demystify these models through input attribution and neuron role analysis, there's been a notable gap in considering layer-level information and the holistic path of information flow across layers. In this paper, we investigate the significance of influential neuron paths within vision Transformers, which is a path of neurons from the model input to output that impacts the model inference most significantly. We first propose a joint influence measure to assess the contribution of a set of neurons to the model outcome. And we further provide a layer-progressive neuron locating approach that efficiently selects the most influential neuron at each layer trying to discover the crucial neuron path from input to output within the target model. Our experiments demonstrate the superiority of our method finding the most influential neuron path along which the information flows, over the existing baseline solutions. Additionally, the neuron paths have illustrated that vision Transformers exhibit some specific inner working mechanism for processing the visual information within the same image category. We further analyze the key effects of these neurons on the image classification task, showcasing that the found neuron paths have already preserved the model capability on downstream tasks, which may also shed some lights on real-world applications like model pruning. The project website including implementation code is available at https://foundation-model-research.github.io/NeuronPath/.

Yi Wang, Hongye Qiu, Yue Xu et al.

Large Language Models (LLMs) and Vision Language Models (VLMs) have demonstrated impressive capabilities but remain vulnerable to jailbreaking attacks, where adversaries exploit textual or visual triggers to bypass safety guardrails. Recent defenses typically rely on safety fine-tuning or external filters to reduce the model's likelihood of producing harmful content. While effective to some extent, these methods often incur significant computational overheads and suffer from the safety utility trade-off, degrading the model's performance on benign tasks. To address these challenges, we propose EVA (Editing for Versatile Alignment against Jailbreaks), a novel framework that pioneers the application of direct model editing for safety alignment. EVA reframes safety alignment as a precise knowledge correction task. Instead of retraining massive parameters, EVA identifies and surgically edits specific neurons responsible for the model's susceptibility to harmful instructions, while leaving the vast majority of the model unchanged. By localizing the updates, EVA effectively neutralizes harmful behaviors without compromising the model's general reasoning capabilities. Extensive experiments demonstrate that EVA outperforms baselines in mitigating jailbreaks across both LLMs and VLMs, offering a precise and efficient solution for post-deployment safety alignment.

Yue Xu, Chengyan Fu, Li Xiong et al.

Pre-training large language models (LLMs) on vast text corpora enhances natural language processing capabilities but risks encoding social biases, particularly gender bias. While parameter-modification methods like fine-tuning mitigate bias, they are resource-intensive, unsuitable for closed-source models, and lack adaptability to evolving societal norms. Instruction-based approaches offer flexibility but often compromise task performance. To address these limitations, we propose $\textbf{FaIRMaker}$, an automated and model-independent framework that employs an $\textbf{auto-search and refinement}$ paradigm to adaptively generate Fairwords, which act as instructions integrated into input queries to reduce gender bias and enhance response quality. Extensive experiments demonstrate that FaIRMaker automatically searches for and dynamically refines Fairwords, effectively mitigating gender bias while preserving task integrity and ensuring compatibility with both API-based and open-source LLMs.

Hanzhuo Huang, Qingyang Bao, Zekai Gu et al.

In this paper, we propose a 3D asset-referenced diffusion model for image generation, exploring how to integrate 3D assets into image diffusion models. Existing reference-based image generation methods leverage large-scale pretrained diffusion models and demonstrate strong capability in generating diverse images conditioned on a single reference image. However, these methods are limited to single-image references and cannot leverage 3D assets, constraining their practical versatility. To address this gap, we present a cross-domain diffusion model with dual-branch perception that leverages multi-view RGB images and point maps of 3D assets to jointly model their colors and canonical-space coordinates, achieving precise consistency between generated images and the 3D references. Our spatially aligned dual-branch generation architecture and domain-decoupled generation mechanism ensure the simultaneous generation of two spatially aligned but content-disentangled outputs, RGB images and point maps, linking 2D image attributes with 3D asset attributes. Experiments show that our approach effectively uses 3D assets as references to produce images consistent with the given assets, opening new possibilities for combining diffusion models with 3D content creation.

Yukai Zhou, Sibei Yang, Wenjie Wang

Large language models (LLMs) are increasingly deployed in real-world applications, raising concerns about their security. While jailbreak attacks highlight failures under overtly harmful queries, they overlook a critical risk: incorrectly answering harmless-looking inputs can be dangerous and cause real-world harm (Implicit Harm). We systematically reformulate the LLM risk landscape through a structured quadrant perspective based on output factuality and input harmlessness, uncovering an overlooked high-risk region. To investigate this gap, we propose JailFlipBench, a benchmark aims to capture implicit harm, spanning single-modal, multimodal, and factual extension scenarios with diverse evaluation metrics. We further develop initial JailFlip attack methodologies and conduct comprehensive evaluations across multiple open-source and black-box LLMs, show that implicit harm present immediate and urgent real-world risks, calling for broader LLM safety assessments and alignment beyond conventional jailbreak paradigms.

Sibei Yang, Guanbin Li, Yizhou Yu

Grounding referring expressions aims to locate in an image an object referred to by a natural language expression. The linguistic structure of a referring expression provides a layout of reasoning over the visual contents, and it is often crucial to align and jointly understand the image and the referring expression. In this paper, we propose a scene graph guided modular network (SGMN), which performs reasoning over a semantic graph and a scene graph with neural modules under the guidance of the linguistic structure of the expression. In particular, we model the image as a structured semantic graph, and parse the expression into a language scene graph. The language scene graph not only decodes the linguistic structure of the expression, but also has a consistent representation with the image semantic graph. In addition to exploring structured solutions to grounding referring expressions, we also propose Ref-Reasoning, a large-scale real-world dataset for structured referring expression reasoning. We automatically generate referring expressions over the scene graphs of images using diverse expression templates and functional programs. This dataset is equipped with real-world visual contents as well as semantically rich expressions with different reasoning layouts. Experimental results show that our SGMN not only significantly outperforms existing state-of-the-art algorithms on the new Ref-Reasoning dataset, but also surpasses state-of-the-art structured methods on commonly used benchmark datasets. It can also provide interpretable visual evidences of reasoning. Data and code are available at https://github.com/sibeiyang/sgmn

Sibei Yang, Guanbin Li, Yizhou Yu

Grounding referring expressions in images aims to locate the object instance in an image described by a referring expression. It involves a joint understanding of natural language and image content, and is essential for a range of visual tasks related to human-computer interaction. As a language-to-vision matching task, the core of this problem is to not only extract all the necessary information (i.e., objects and the relationships among them) in both the image and referring expression, but also make full use of context information to align cross-modal semantic concepts in the extracted information. Unfortunately, existing work on grounding referring expressions fails to accurately extract multi-order relationships from the referring expression and associate them with the objects and their related contexts in the image. In this paper, we propose a Cross-Modal Relationship Extractor (CMRE) to adaptively highlight objects and relationships (spatial and semantic relations) related to the given expression with a cross-modal attention mechanism, and represent the extracted information as a language-guided visual relation graph. In addition, we propose a Gated Graph Convolutional Network (GGCN) to compute multimodal semantic contexts by fusing information from different modes and propagating multimodal information in the structured relation graph. Experimental results on three common benchmark datasets show that our Cross-Modal Relationship Inference Network, which consists of CMRE and GGCN, significantly surpasses all existing state-of-the-art methods. Code is available at https://github.com/sibeiyang/sgmn/tree/master/lib/cmrin_models

Han Liang, Jiacheng Bao, Ruichi Zhang et al.

We have recently seen tremendous progress in realistic text-to-motion generation. Yet, the existing methods often fail or produce implausible motions with unseen text inputs, which limits the applications. In this paper, we present OMG, a novel framework, which enables compelling motion generation from zero-shot open-vocabulary text prompts. Our key idea is to carefully tailor the pretrain-then-finetune paradigm into the text-to-motion generation. At the pre-training stage, our model improves the generation ability by learning the rich out-of-domain inherent motion traits. To this end, we scale up a large unconditional diffusion model up to 1B parameters, so as to utilize the massive unlabeled motion data up to over 20M motion instances. At the subsequent fine-tuning stage, we introduce motion ControlNet, which incorporates text prompts as conditioning information, through a trainable copy of the pre-trained model and the proposed novel Mixture-of-Controllers (MoC) block. MoC block adaptively recognizes various ranges of the sub-motions with a cross-attention mechanism and processes them separately with the text-token-specific experts. Such a design effectively aligns the CLIP token embeddings of text prompts to various ranges of compact and expressive motion features. Extensive experiments demonstrate that our OMG achieves significant improvements over the state-of-the-art methods on zero-shot text-to-motion generation. Project page: https://tr3e.github.io/omg-page.

Yumeng Liu, Yaxun Yang, Youzhuo Wang et al.

In this paper, we introduce RealDex, a pioneering dataset capturing authentic dexterous hand grasping motions infused with human behavioral patterns, enriched by multi-view and multimodal visual data. Utilizing a teleoperation system, we seamlessly synchronize human-robot hand poses in real time. This collection of human-like motions is crucial for training dexterous hands to mimic human movements more naturally and precisely. RealDex holds immense promise in advancing humanoid robot for automated perception, cognition, and manipulation in real-world scenarios. Moreover, we introduce a cutting-edge dexterous grasping motion generation framework, which aligns with human experience and enhances real-world applicability through effectively utilizing Multimodal Large Language Models. Extensive experiments have demonstrated the superior performance of our method on RealDex and other open datasets. The complete dataset and code will be made available upon the publication of this work.

Qiyuan Dai, Sibei Yang

Referring image segmentation (RIS) aims to precisely segment referents in images through corresponding natural language expressions, yet relying on cost-intensive mask annotations. Weakly supervised RIS thus learns from image-text pairs to pixel-level semantics, which is challenging for segmenting fine-grained masks. A natural approach to enhancing segmentation precision is to empower weakly supervised RIS with the image segmentation foundation model SAM. Nevertheless, we observe that simply integrating SAM yields limited benefits and can even lead to performance regression due to the inevitable noise issues and challenges in excessive focus on object parts. In this paper, we present an innovative framework, Point PrompTing (PPT), incorporated with the proposed multi-source curriculum learning strategy to address these challenges. Specifically, the core of PPT is a point generator that not only harnesses CLIP's text-image alignment capability and SAM's powerful mask generation ability but also generates negative point prompts to address the noisy and excessive focus issues inherently and effectively. In addition, we introduce a curriculum learning strategy with object-centric images to help PPT gradually learn from simpler yet precise semantic alignment to more complex RIS. Experiments demonstrate that our PPT significantly and consistently outperforms prior weakly supervised techniques on mIoU by 11.34%, 14.14%, and 6.97% across RefCOCO, RefCOCO+, and G-Ref, respectively.

Hanzhuo Huang, Yuan Liu, Ge Zheng et al.

In this paper, we present MVTokenFlow for high-quality 4D content creation from monocular videos. Recent advancements in generative models such as video diffusion models and multiview diffusion models enable us to create videos or 3D models. However, extending these generative models for dynamic 4D content creation is still a challenging task that requires the generated content to be consistent spatially and temporally. To address this challenge, MVTokenFlow utilizes the multiview diffusion model to generate multiview images on different timesteps, which attains spatial consistency across different viewpoints and allows us to reconstruct a reasonable coarse 4D field. Then, MVTokenFlow further regenerates all the multiview images using the rendered 2D flows as guidance. The 2D flows effectively associate pixels from different timesteps and improve the temporal consistency by reusing tokens in the regeneration process. Finally, the regenerated images are spatiotemporally consistent and utilized to refine the coarse 4D field to get a high-quality 4D field. Experiments demonstrate the effectiveness of our design and show significantly improved quality than baseline methods.

Chunlin Yu, Hanqing Wang, Ye Shi et al.

3D affordance segmentation aims to link human instructions to touchable regions of 3D objects for embodied manipulations. Existing efforts typically adhere to single-object, single-affordance paradigms, where each affordance type or explicit instruction strictly corresponds to a specific affordance region and are unable to handle long-horizon tasks. Such a paradigm cannot actively reason about complex user intentions that often imply sequential affordances. In this paper, we introduce the Sequential 3D Affordance Reasoning task, which extends the traditional paradigm by reasoning from cumbersome user intentions and then decomposing them into a series of segmentation maps. Toward this, we construct the first instruction-based affordance segmentation benchmark that includes reasoning over both single and sequential affordances, comprising 180K instruction-point cloud pairs. Based on the benchmark, we propose our model, SeqAfford, to unlock the 3D multi-modal large language model with additional affordance segmentation abilities, which ensures reasoning with world knowledge and fine-grained affordance grounding in a cohesive framework. We further introduce a multi-granular language-point integration module to endow 3D dense prediction. Extensive experimental evaluations show that our model excels over well-established methods and exhibits open-world generalization with sequential reasoning abilities.

Zijian He, Yuwei Ning, Yipeng Qin et al.

Virtual Try-On (VTON) is a transformative technology in e-commerce and fashion design, enabling realistic digital visualization of clothing on individuals. In this work, we propose VTON 360, a novel 3D VTON method that addresses the open challenge of achieving high-fidelity VTON that supports any-view rendering. Specifically, we leverage the equivalence between a 3D model and its rendered multi-view 2D images, and reformulate 3D VTON as an extension of 2D VTON that ensures 3D consistent results across multiple views. To achieve this, we extend 2D VTON models to include multi-view garments and clothing-agnostic human body images as input, and propose several novel techniques to enhance them, including: i) a pseudo-3D pose representation using normal maps derived from the SMPL-X 3D human model, ii) a multi-view spatial attention mechanism that models the correlations between features from different viewing angles, and iii) a multi-view CLIP embedding that enhances the garment CLIP features used in 2D VTON with camera information. Extensive experiments on large-scale real datasets and clothing images from e-commerce platforms demonstrate the effectiveness of our approach. Project page: https://scnuhealthy.github.io/VTON360.

Qiyuan Dai, Sibei Yang

Vision-Language Models (VLMs) have become prominent in open-world image recognition for their strong generalization abilities. Yet, their effectiveness in practical applications is compromised by domain shifts and distributional changes, especially when test data distributions diverge from training data. Therefore, the paradigm of test-time adaptation (TTA) has emerged, enabling the use of online off-the-shelf data at test time, supporting independent sample predictions, and eliminating reliance on test annotations. Traditional TTA methods, however, often rely on costly training or optimization processes, or make unrealistic assumptions about accessing or storing historical training and test data. Instead, this study proposes FreeTTA, a training-free and universally available method that makes no assumptions, to enhance the flexibility of TTA. More importantly, FreeTTA is the first to explicitly model the test data distribution, enabling the use of intrinsic relationships among test samples to enhance predictions of individual samples without simultaneous access--a direction not previously explored. FreeTTA achieves these advantages by introducing an online EM algorithm that utilizes zero-shot predictions from VLMs as priors to iteratively compute the posterior probabilities of each online test sample and update parameters. Experiments demonstrate that FreeTTA achieves stable and significant improvements compared to state-of-the-art methods across 15 datasets in both cross-domain and out-of-distribution settings.

Yingdong Shi, Changming Li, Yifan Wang et al.

Diffusion models have demonstrated impressive capabilities in synthesizing diverse content. However, despite their high-quality outputs, these models often perpetuate social biases, including those related to gender and race. These biases can potentially contribute to harmful real-world consequences, reinforcing stereotypes and exacerbating inequalities in various social contexts. While existing research on diffusion bias mitigation has predominantly focused on guiding content generation, it often neglects the intrinsic mechanisms within diffusion models that causally drive biased outputs. In this paper, we investigate the internal processes of diffusion models, identifying specific decision-making mechanisms, termed bias features, embedded within the model architecture. By directly manipulating these features, our method precisely isolates and adjusts the elements responsible for bias generation, permitting granular control over the bias levels in the generated content. Through experiments on both unconditional and conditional diffusion models across various social bias attributes, we demonstrate our method's efficacy in managing generation distribution while preserving image quality. We also dissect the discovered model mechanism, revealing different intrinsic features controlling fine-grained aspects of generation, boosting further research on mechanistic interpretability of diffusion models.

Qiyuan Dai, Hanzhuo Huang, Yu Wu et al.

Generalized Category Discovery (GCD) aims to recognize unlabeled images from known and novel classes by distinguishing novel classes from known ones, while also transferring knowledge from another set of labeled images with known classes. Existing GCD methods rely on self-supervised vision transformers such as DINO for representation learning. However, focusing solely on the global representation of the DINO CLS token introduces an inherent trade-off between discriminability and generalization. In this paper, we introduce an adaptive part discovery and learning method, called APL, which generates consistent object parts and their correspondences across different similar images using a set of shared learnable part queries and DINO part priors, without requiring any additional annotations. More importantly, we propose a novel all-min contrastive loss to learn discriminative yet generalizable part representation, which adaptively highlights discriminative object parts to distinguish similar categories for enhanced discriminability while simultaneously sharing other parts to facilitate knowledge transfer for improved generalization. Our APL can easily be incorporated into different GCD frameworks by replacing their CLS token feature with our part representations, showing significant enhancements on fine-grained datasets.

Yi Wang, Fenghua Weng, Sibei Yang et al.

Large Language Models (LLMs) are widely applied in decision making, but their deployment is threatened by jailbreak attacks, where adversarial users manipulate model behavior to bypass safety measures. Existing defense mechanisms, such as safety fine-tuning and model editing, either require extensive parameter modifications or lack precision, leading to performance degradation on general tasks, which is unsuitable to post-deployment safety alignment. To address these challenges, we propose DELMAN (Dynamic Editing for LLMs JAilbreak DefeNse), a novel approach leveraging direct model editing for precise, dynamic protection against jailbreak attacks. DELMAN directly updates a minimal set of relevant parameters to neutralize harmful behaviors while preserving the model's utility. To avoid triggering a safe response in benign context, we incorporate KL-divergence regularization to ensure the updated model remains consistent with the original model when processing benign queries. Experimental results demonstrate that DELMAN outperforms baseline methods in mitigating jailbreak attacks while preserving the model's utility, and adapts seamlessly to new attack instances, providing a practical and efficient solution for post-deployment model protection.

Ge Zheng, Jiaye Qian, Jiajin Tang et al.

Large Vision-Language Models (LVLMs) have made significant progress in recent years but are also prone to hallucination issues. They exhibit more hallucinations in longer, free-form responses, often attributed to accumulated uncertainties. In this paper, we ask: Does increased hallucination result solely from length-induced errors, or is there a deeper underlying mechanism? After a series of preliminary experiments and findings, we suggest that the risk of hallucinations is not caused by length itself but by the increased reliance on context for coherence and completeness in longer responses. Building on these insights, we propose a novel "induce-detect-suppress" framework that actively induces hallucinations through deliberately designed contexts, leverages induced instances for early detection of high-risk cases, and ultimately suppresses potential object-level hallucinations during actual decoding. Our approach achieves consistent, significant improvements across all benchmarks, demonstrating its efficacy. The strong detection and improved hallucination mitigation not only validate our framework but, more importantly, re-validate our hypothesis on context. Rather than solely pursuing performance gains, this study aims to provide new insights and serves as a first step toward a deeper exploration of hallucinations in LVLMs' longer responses.

Jiajin Tang, Zhengxuan Wei, Yuchen Zhu et al.

Temporal sentence grounding aims to identify exact moments in a video that correspond to a given textual query, typically addressed with detection transformer (DETR) solutions. However, we find that typical strategies designed to enhance DETR do not improve, and may even degrade, its performance in this task. We systematically analyze and identify the root causes of this abnormal behavior: (1) conflicts between queries from similar target moments and (2) internal query conflicts due to the tension between global semantics and local localization. Building on these insights, we propose a simple yet powerful baseline, Sim-DETR, which extends the standard DETR with two minor modifications in the decoder layers: (1) constraining self-attention between queries based on their semantic and positional overlap and (2) adding query-to-frame alignment to bridge the global and local contexts. Experiments demonstrate that Sim-DETR unlocks the full potential of DETR for temporal sentence grounding, offering a strong baseline for future research.

Jiaye Qian, Ge Zheng, Yuchen Zhu et al.

Despite their impressive performance across a wide range of tasks, Large Vision-Language Models (LVLMs) remain prone to hallucination. In this study, we propose a comprehensive intervention framework aligned with the transformer's causal architecture in LVLMs, integrating the effects of different intervention paths on hallucination. We find that hallucinations in LVLMs do not arise from a single causal path, but rather from the interplay among image-to-input-text, image-to-output-text, and text-to-text pathways. For the first time, we also find that LVLMs rely on different pathways depending on the question-answer alignment format. Building on these insights, we propose simple yet effective methods to identify and intervene on critical hallucination heads within each pathway, tailored to discriminative and generative formats. Experiments across multiple benchmarks demonstrate that our approach consistently reduces hallucinations across diverse alignment types.

Zhengxuan Wei, Jiajin Tang, Sibei Yang

Existing Moment Retrieval methods face three critical bottlenecks: (1) data scarcity forces models into shallow keyword-feature associations; (2) boundary ambiguity in transition regions between adjacent events; (3) insufficient discrimination of fine-grained semantics (e.g., distinguishing ``kicking" vs. ``throwing" a ball). In this paper, we propose a zero-external-dependency Augmented Moment Retrieval framework, AMR, designed to overcome local optima caused by insufficient data annotations and the lack of robust boundary and semantic discrimination capabilities. AMR is built upon two key insights: (1) it resolves ambiguous boundary information and semantic confusion in existing annotations without additional data (avoiding costly manual labeling), and (2) it preserves boundary and semantic discriminative capabilities enhanced by training while generalizing to real-world scenarios, significantly improving performance. Furthermore, we propose a two-stage training framework with cold-start and distillation adaptation. The cold-start stage employs curriculum learning on augmented data to build foundational boundary/semantic awareness. The distillation stage introduces dual query sets: Original Queries maintain DETR-based localization using frozen Base Queries from the cold-start model, while Active Queries dynamically adapt to real-data distributions. A cross-stage distillation loss enforces consistency between Original and Base Queries, preventing knowledge forgetting while enabling real-world generalization. Experiments on multiple benchmarks show that AMR achieves improved performance over prior state-of-the-art approaches.

Jiajin Tang, Zhengxuan Wei, Ge Zheng et al.

Humans can perform previously unexperienced interactions with novel objects simply by observing others engage with them. Weakly-supervised affordance grounding mimics this process by learning to locate object regions that enable actions on egocentric images, using exocentric interaction images with image-level annotations. However, extracting affordance knowledge solely from exocentric images and transferring it one-way to egocentric images limits the applicability of previous works in complex interaction scenarios. Instead, this study introduces LoopTrans, a novel closed-loop framework that not only transfers knowledge from exocentric to egocentric but also transfers back to enhance exocentric knowledge extraction. Within LoopTrans, several innovative mechanisms are introduced, including unified cross-modal localization and denoising knowledge distillation, to bridge domain gaps between object-centered egocentric and interaction-centered exocentric images while enhancing knowledge transfer. Experiments show that LoopTrans achieves consistent improvements across all metrics on image and video benchmarks, even handling challenging scenarios where object interaction regions are fully occluded by the human body.

Yulin Zhang, Cheng Shi, Yang Wang et al.

Envision an AI capable of functioning in human-like settings, moving beyond mere observation to actively understand, anticipate, and proactively respond to unfolding events. Towards this vision, we focus on the innovative task where, given ego-streaming video input, an assistant proactively answers diverse, evolving questions at the opportune moment, while maintaining synchronized perception and reasoning. This task embodies three key properties: (1) Proactive Coherence, (2) Just-in-Time Responsiveness, and (3) Synchronized Efficiency. To evaluate and address these properties, we first introduce ESTP-Bench (Ego Streaming Proactive Benchmark) alongside the ESTP-F1 metric-a novel framework designed for their rigorous assessment. Secondly, we propose a comprehensive technical pipeline to enable models to tackle this challenging task. This pipeline comprises: (1) a data engine, (2) a multi-stage training strategy, and (3) a proactive dynamic compression technique. Our proposed model effectively addresses these critical properties while outperforming multiple baselines across diverse online and offline benchmarks. Project Page:https://zhangyl4.github.io/publications/eyes-wide-open/

Cheng Shi, Yizhou Yu, Sibei Yang

This study identifies that visual-related functional decoding is distributed across different decoder layers in Multimodal Large Language Models (MLLMs). Typically, each function, such as counting, grounding, or OCR recognition, narrows down to two or three layers, which we define as Vision Function Layers (VFL). Additionally, the depth and its order of different VFLs exhibits a consistent pattern across different MLLMs, which is well-aligned with human behaviors (e.g., recognition occurs first, followed by counting, and then grounding). These findings are derived from Visual Token Swapping, our novel analytical framework that modifies targeted KV cache entries to precisely elucidate layer-specific functions during decoding. Furthermore, these insights offer substantial utility in tailoring MLLMs for real-world downstream applications. For instance, when LoRA training is selectively applied to VFLs whose functions align with the training data, VFL-LoRA not only outperform full-LoRA but also prevent out-of-domain function forgetting. Moreover, by analyzing the performance differential on training data when particular VFLs are ablated, VFL-select automatically classifies data by function, enabling highly efficient data selection to directly bolster corresponding capabilities. Consequently, VFL-select surpasses human experts in data selection, and achieves 98% of full-data performance with only 20% of the original dataset. This study delivers deeper comprehension of MLLM visual processing, fostering the creation of more efficient, interpretable, and robust models.

Haoyang Xu, Tianhao Zhao, Sibei Yang et al.

Diffusion models have emerged as a powerful technique for text-to-image (T2I) generation, creating high-quality, diverse images across various domains. However, a common limitation in these models is the incomplete display of objects, where fragments or missing parts undermine the model's performance in downstream applications. In this study, we conduct an in-depth analysis of the incompleteness issue and reveal that the primary factor behind incomplete object generation is the usage of RandomCrop during model training. This widely used data augmentation method, though enhances model generalization ability, disrupts object continuity during training. To address this, we propose a training-free solution that penalizes activation values at image boundaries during the early denoising steps. Our method is easily applicable to pre-trained Stable Diffusion models with minimal modifications and negligible computational overhead. Extensive experiments demonstrate the effectiveness of our method, showing substantial improvements in object integrity and image quality.

Bin Yang, Yulin Zhang, Hong-Yu Zhou et al.

Detection transformers have been applied to human-object interaction (HOI) detection, enhancing the localization and recognition of human-action-object triplets in images. Despite remarkable progress, this study identifies a critical issue-"Toxic Siblings" bias-which hinders the interaction decoder's learning, as numerous similar yet distinct HOI triplets interfere with and even compete against each other both input side and output side to the interaction decoder. This bias arises from high confusion among sibling triplets/categories, where increased similarity paradoxically reduces precision, as one's gain comes at the expense of its toxic sibling's decline. To address this, we propose two novel debiasing learning objectives-"contrastive-then-calibration" and "merge-then-split"-targeting the input and output perspectives, respectively. The former samples sibling-like incorrect HOI triplets and reconstructs them into correct ones, guided by strong positional priors. The latter first learns shared features among sibling categories to distinguish them from other groups, then explicitly refines intra-group differentiation to preserve uniqueness. Experiments show that we significantly outperform both the baseline (+9.18% mAP on HICO-Det) and the state-of-the-art (+3.59% mAP) across various settings.

Shuo Xu, Sai Wang, Xinyue Hu et al.

Compositional Zero-Shot Learning (CZSL) aims to learn semantic primitives (attributes and objects) from seen compositions and recognize unseen attribute-object compositions. Existing CZSL datasets focus on single attributes, neglecting the fact that objects naturally exhibit multiple interrelated attributes. Their narrow attribute scope and single attribute labeling introduce annotation biases, misleading the learning of attributes and causing inaccurate evaluation. To address these issues, we introduce the Multi-Attribute Composition (MAC) dataset, encompassing 22,838 images and 17,627 compositions with comprehensive and representative attribute annotations. MAC shows complex relationship between attributes and objects, with each attribute type linked to an average of 82.2 object types, and each object type associated with 31.4 attribute types. Based on MAC, we propose multi-attribute compositional zero-shot learning that requires deeper semantic understanding and advanced attribute associations, establishing a more realistic and challenging benchmark for CZSL. We also propose Multi-attribute Visual-Primitive Integrator (MVP-Integrator), a robust baseline for multi-attribute CZSL, which disentangles semantic primitives and performs effective visual-primitive association. Experimental results demonstrate that MVP-Integrator significantly outperforms existing CZSL methods on MAC with improved inference efficiency.

Cheng Shi, Sibei Yang

Prompt engineering is a powerful tool used to enhance the performance of pre-trained models on downstream tasks. For example, providing the prompt "Let's think step by step" improved GPT-3's reasoning accuracy to 63% on MutiArith while prompting "a photo of" filled with a class name enables CLIP to achieve $80$\% zero-shot accuracy on ImageNet. While previous research has explored prompt learning for the visual modality, analyzing what constitutes a good visual prompt specifically for image recognition is limited. In addition, existing visual prompt tuning methods' generalization ability is worse than text-only prompting tuning. This paper explores our key insight: synthetic text images are good visual prompts for vision-language models! To achieve that, we propose our LoGoPrompt, which reformulates the classification objective to the visual prompt selection and addresses the chicken-and-egg challenge of first adding synthetic text images as class-wise visual prompts or predicting the class first. Without any trainable visual prompt parameters, experimental results on 16 datasets demonstrate that our method consistently outperforms state-of-the-art methods in few-shot learning, base-to-new generalization, and domain generalization.

Cheng Shi, Sibei Yang

Vision-language models such as CLIP have boosted the performance of open-vocabulary object detection, where the detector is trained on base categories but required to detect novel categories. Existing methods leverage CLIP's strong zero-shot recognition ability to align object-level embeddings with textual embeddings of categories. However, we observe that using CLIP for object-level alignment results in overfitting to base categories, i.e., novel categories most similar to base categories have particularly poor performance as they are recognized as similar base categories. In this paper, we first identify that the loss of critical fine-grained local image semantics hinders existing methods from attaining strong base-to-novel generalization. Then, we propose Early Dense Alignment (EDA) to bridge the gap between generalizable local semantics and object-level prediction. In EDA, we use object-level supervision to learn the dense-level rather than object-level alignment to maintain the local fine-grained semantics. Extensive experiments demonstrate our superior performance to competing approaches under the same strict setting and without using external training resources, i.e., improving the +8.4% novel box AP50 on COCO and +3.9% rare mask AP on LVIS.

Jiajin Tang, Ge Zheng, Jingyi Yu et al.

Task driven object detection aims to detect object instances suitable for affording a task in an image. Its challenge lies in object categories available for the task being too diverse to be limited to a closed set of object vocabulary for traditional object detection. Simply mapping categories and visual features of common objects to the task cannot address the challenge. In this paper, we propose to explore fundamental affordances rather than object categories, i.e., common attributes that enable different objects to accomplish the same task. Moreover, we propose a novel multi-level chain-of-thought prompting (MLCoT) to extract the affordance knowledge from large language models, which contains multi-level reasoning steps from task to object examples to essential visual attributes with rationales. Furthermore, to fully exploit knowledge to benefit object recognition and localization, we propose a knowledge-conditional detection framework, namely CoTDet. It conditions the detector from the knowledge to generate object queries and regress boxes. Experimental results demonstrate that our CoTDet outperforms state-of-the-art methods consistently and significantly (+15.6 box AP and +14.8 mask AP) and can generate rationales for why objects are detected to afford the task.

Cheng Shi, Sibei Yang

Embodied Reference Understanding studies the reference understanding in an embodied fashion, where a receiver is required to locate a target object referred to by both language and gesture of the sender in a shared physical environment. Its main challenge lies in how to make the receiver with the egocentric view access spatial and visual information relative to the sender to judge how objects are oriented around and seen from the sender, i.e., spatial and visual perspective-taking. In this paper, we propose a REasoning from your Perspective (REP) method to tackle the challenge by modeling relations between the receiver and the sender and the sender and the objects via the proposed novel view rotation and relation reasoning. Specifically, view rotation first rotates the receiver to the position of the sender by constructing an embodied 3D coordinate system with the position of the sender as the origin. Then, it changes the orientation of the receiver to the orientation of the sender by encoding the body orientation and gesture of the sender. Relation reasoning models the nonverbal and verbal relations between the sender and the objects by multi-modal cooperative reasoning in gesture, language, visual content, and spatial position. Experiment results demonstrate the effectiveness of REP, which consistently surpasses all existing state-of-the-art algorithms by a large margin, i.e., +5.22% absolute accuracy in terms of Prec0.5 on YouRefIt.

Jiajin Tang, Ge Zheng, Cheng Shi et al.

Referring image segmentation aims to segment the target referent in an image conditioning on a natural language expression. Existing one-stage methods employ per-pixel classification frameworks, which attempt straightforwardly to align vision and language at the pixel level, thus failing to capture critical object-level information. In this paper, we propose a mask classification framework, Contrastive Grouping with Transformer network (CGFormer), which explicitly captures object-level information via token-based querying and grouping strategy. Specifically, CGFormer first introduces learnable query tokens to represent objects and then alternately queries linguistic features and groups visual features into the query tokens for object-aware cross-modal reasoning. In addition, CGFormer achieves cross-level interaction by jointly updating the query tokens and decoding masks in every two consecutive layers. Finally, CGFormer cooperates contrastive learning to the grouping strategy to identify the token and its mask corresponding to the referent. Experimental results demonstrate that CGFormer outperforms state-of-the-art methods in both segmentation and generalization settings consistently and significantly.

Hong-Yu Zhou, Chixiang Lu, Sibei Yang et al.

Preserving maximal information is one of principles of designing self-supervised learning methodologies. To reach this goal, contrastive learning adopts an implicit way which is contrasting image pairs. However, we believe it is not fully optimal to simply use the contrastive estimation for preservation. Moreover, it is necessary and complemental to introduce an explicit solution to preserve more information. From this perspective, we introduce Preservational Learning to reconstruct diverse image contexts in order to preserve more information in learned representations. Together with the contrastive loss, we present Preservational Contrastive Representation Learning (PCRL) for learning self-supervised medical representations. PCRL provides very competitive results under the pretraining-finetuning protocol, outperforming both self-supervised and supervised counterparts in 5 classification/segmentation tasks substantially.