论文检索

Joonmyung Choi, Sanghyeok Lee, Jongha Kim, Sehyung Kim, Dohwan Ko, Jihyung Kil, Hyunwoo J. Kim

Recent advances in vision-language models have shown strong performance across diverse multimodal tasks, including document question answering that leverages structured visual cues from text, tables, and figures. However, unlike natural images, document images contain large backgrounds and only sparse supporting evidence, leading to the waste of substantial computational resources, especially for long documents. We observe that existing token reduction methods for natural images and videos fall short in utilizing the structural sparsity unique to documents. To address this, we propose DOCPRUNE, a training-free document token pruning framework designed for efficient long document understanding. The proposed method preserves only the essential tokens for the task while removing unnecessary ones, such as background or question-irrelevant tokens. Moreover, it automatically selects the appropriate layers to initiate token pruning based on the model's level of comprehension. Our experiments on the M3DocRAG benchmark show that DOCPRUNE improves throughput by 3.0x and 3.3x in the encoder and decoder, respectively, while boosting the F1 score by +1.0, achieving both higher accuracy and efficiency without any additional training.

Chubin Chen, Sujie Hu, Jiashu Zhu, Meiqi Wu, Jintao Chen, Yanxun Li, Nisha Huang, Chengyu Fang, Jiahong Wu, Xiangxiang Chu 等

Recent studies have demonstrated significant progress in aligning text-to-image diffusion models with human preference via Reinforcement Learning from Human Feedback. However, while existing methods achieve high scores on automated reward metrics, they often lead to Preference Mode Collapse (PMC)-a specific form of reward hacking where models converge on narrow, high-scoring outputs (e.g., images with monolithic styles or pervasive overexposure), severely degrading generative diversity. In this work, we introduce and quantify this phenomenon, proposing DivGenBench, a novel benchmark designed to measure the extent of PMC. We posit that this collapse is driven by over-optimization along the reward model's inherent biases. Building on this analysis, we propose Directional Decoupling Alignment (D^2-Align), a novel framework that mitigates PMC by directionally correcting the reward signal.Specifically, our method first learns a directional correction within the reward model's embedding space while keeping the model frozen.This correction is then applied to the reward signal during the optimization process, preventing the model from collapsing into specific modes and thereby maintaining diversity. Our comprehensive evaluation, combining qualitative analysis with quantitative metrics for both quality and diversity, reveals that D^2-Align achieves superior alignment with human preference.

Ling Wang, Yunfan Lu, Wenzong Ma, Huizai Yao, Pengteng Li, Hui Xiong

Clear imaging under hazy conditions is a critical task. Prior-based and neural methods have improved results. However, they operate on RGB frames, which suffer from limited dynamic range. Therefore, dehazing remains ill- posed and can erase structure and illumination details. To address this, we use event cameras for dehazing for the first time. Event cameras offer much higher HDR (120dB vs. 60dB) and microsecond latency, therefore they suit hazy scenes. In practice, transferring HDR cues from events to frames is hard because real paired data are scarce. To tackle this, we propose an event-guided diffusion model that utilizes the strong generative priors of diffusion mod- els to reconstruct clear images from hazy inputs by effec- tively transferring HDR information from events. Specifi- cally, we design an event-guided module that maps sparse HDR event features, e.g., edges, corners, into the diffusion latent space. This clear conditioning provides precise struc- tural guidance during generation, improves visual realism, and reduces semantic drift. For real-world evaluation, we collect a drone dataset in heavy haze (AQI = 341) with synchronized RGB and event sensors. Experiments on two benchmarks and our dataset achieve state-of-the-art results.

Shezheng Song, Shasha Li, Shan Zhao, Xiaopeng Li, Qian Wan, Chengyu Wang, Tianwei Yan, Ma Jun, Jie Yu

Multimodal Large Language Models (MLLMs) have achieved remarkable progress in vision-language understanding, yet how they internally integrate visual and textual information remains poorly understood. To bridge this gap, we perform a systematic layer-wise masking analysis across multiple architectures, revealing how visual-text fusion evolves within MLLMs. The results show that fusion emerges at several specific layers rather than being uniformly distributed across the network, and certain models exhibit a late-stage "review" phenomenon where visual signals are reactivated before output generation. Besides, we further analyze layer-wise attention evolution and observe persistent high-attention noise on irrelevant regions, along with gradually increasing attention on text-aligned areas. Guided by these insights, we introduce a training-free contrastive attention framework that models the transformation between early fusion and final layers to highlight meaningful attention shifts. Extensive experiments across various MLLMs and benchmarks validate our analysis and demonstrate that the proposed approach improves multimodal reasoning performance.Code will be released.

Xilin He, Xiaole Xian, Xiangyu Yue, Muhammad Haris Khan

Style generation has made significant progress through diffusion models. Recent efforts have explored reinforcement learning with human-preference reward models to enhance diffusion models for general downstream applications. However, we identify a critical limitation: existing human-preference reward models struggle to effectively perceive image style, resulting in suboptimal performance after reinforcement fine-tuning. To address this, we first introduce a large-scale style reward modeling dataset comprising 400K paired samples spanning 1,000 diverse style categories, augmented with textual instructions and style reward annotations. We then propose StyleDoctor, a novel style perception reward model capable of jointly evaluating style consistency between paired images and style-text alignment. StyleDoctor outperforms existing style perception models in both style retrieval and generation tasks. Extensive quantitative and qualitative experiments demonstrate the superiority of StyleDoctor over competing approaches, showcasing its efficiency and versatility in style-conditioned generation. Dataset and code are available at StyleDoctor.

Ziheng Ouyang, Yiren Song, Yaoli Liu, Shihao Zhu, Qibin Hou, Ming-Ming Cheng, Mike Zheng Shou

Previous works have explored various customized generation tasks given a reference image, but they still face limitations in generating consistent fine-grained details. In this paper, our aim is to solve the inconsistency problem of generated images by applying a reference-guided post-editing approach and present our ImageCritic. We first construct a dataset of reference-degraded-target triplets obtained via VLM-based selection and explicit degradation, which effectively simulates the common inaccuracies or inconsistencies observed in existing generation models. Furthermore, building on a thorough examination of the model's attention mechanisms and intrinsic representations, we accordingly devise an attention alignment loss and a detail encoder to precisely rectify inconsistencies. ImageCritic can be integrated into an agent framework to automatically detect inconsistencies and correct them with multi-round and local editing in complex scenarios. Extensive experiments demonstrate that ImageCritic can effectively resolve detail-related issues in various customized generation scenarios, providing significant improvements over existing methods.

Ziliang Chen, Yulu Li, Liangda Fang, Jusheng Zhang, Yongsen Zheng, Quanlong Guan, Xipeng Chen

Open-vocabulary learning on CLIP provides remarkable generalization on diverse concepts, however, falters under the realistic streaming open-world evaluations for Stability against distractor classes and Extensibility to novel classes. Current fine-tuning methods often fail these tests since they are mainly designed for closed-set conditions, leading to the performance gaps while the target vocabulary progressively scales. We formalize a "vocabulary scaling law" showing that these openness measures can be lower-bounded by performance on the full class-name universe, implying that robust fine-tuning should: (i) account for the entire vocabulary, (ii) tune class-name embeddings rather than context, and (iii) enforce orthogonality between prompt embeddings including training and open-set class names. Guided by our analysis, we propose Submodular-Vocabulary Fine-tuning (SVFT), a bi-level optimization framework that approximates the intractable objective of tuning all class name embedding by greedily selecting a small, informative subset of class names via constrained submodular maximization, thus, allows the employment of efficient greedy algorithm for the near-optimal class-name subset selection to fine-tune CLIP instead of using all open classes. Across extensive experiments, SVFT consistently improves both stability and extensibility, advancing the openness and practical robustness of CLIP-based vision-language models.

Jiayu Ding, Xinpeng Liu, Zhiyi Pan, Shiqiang Long, Ge Li

Lifting 2D open-vocabulary understanding into 3D Gaussian Splatting (3DGS) scenes is a critical challenge. Mainstream methods, built on an embedding paradigm, suffer from three key flaws: (i) geometry-semantic inconsistency, where points, rather than objects, serve as the semantic basis, limiting semantic fidelity; (ii) semantic bloat from injecting gigabytes of feature data into the geometry; and (iii) semantic rigidity, as one feature per Gaussian struggles to capture rich polysemy. To overcome these limitations, we introduce ExtrinSplat, a framework built on the extrinsic paradigm that decouples geometry from semantics. Instead of embedding features, ExtrinSplat clusters Gaussians into multi-granularity, overlapping 3D object groups. A Vision-Language Model (VLM) then interprets these groups to generate lightweight textual hypotheses, creating an extrinsic index layer that natively supports complex polysemy. By replacing costly feature embedding with lightweight indices, ExtrinSplat reduces scene adaptation time from hours to minutes and lowers storage overhead by several orders of magnitude. On benchmark tasks for open-vocabulary 3D object selection and semantic segmentation, ExtrinSplat outperforms established embedding-based frameworks, validating the efficacy and efficiency of the proposed extrinsic paradigm.

Xiaogang Du, Jiawei Zhang, Tongfei Liu, Tao Lei, Yingbo Wang

In medical image segmentation tasks, the domain gap caused by the difference in data collection between training and testing data seriously hinders the deployment of pre-trained models in clinical practice. Continual Test-Time Adaptation (CTTA) aims to enable pre-trained models to adapt to continuously changing unlabeled domains, providing an effective approach to solving this problem. However, existing CTTA methods often rely on unreliable supervisory signals, igniting a self-reinforcing cycle of error accumulation that culminates in catastrophic performance degradation. To overcome these challenges, we propose a CTTA via Semantic-Prompt-Enhanced Graph Clustering (SPEGC) for medical image segmentation. First, we design a semantic prompt feature enhancement mechanism that utilizes decoupled commonality and heterogeneity prompt pools to inject global contextual information into local features, alleviating their susceptibility to noise interference under domain shift. Second, based on these enhanced features, we design a differentiable graph clustering solver. This solver reframes global edge sparsification as an optimal transport problem, allowing it to distill a raw similarity matrix into a refined and high-order structural representation in an end-to-end manner. Finally, this robust structural representation is used to guide model adaptation, ensuring predictions are consistent at a cluster-level and dynamically adjusting decision boundaries. Extensive experiments demonstrate that SPEGC outperforms other state-of-the-art CTTA methods on two medical image segmentation benchmarks. The source code is available at https://github.com/Jwei-Z/SPEGC-for-MIS.

Yichen Xie, Chensheng Peng, Mazen Abdelfattah, Yihan Hu, Jiezhi Yang, Eric Higgins, Ryan Brigden, Masayoshi Tomizuka, Wei Zhan

World foundation models aim to simulate the evolution of the real world with physically plausible behavior. Unlike prior methods that handle spatial and temporal correlations separately, we propose RAYNOVA, a geometry-agonistic multiview world model for driving scenarios that employs a dual-causal autoregressive framework. It follows both scale-wise and temporal topological orders in the autoregressive process, and leverages global attention for unified 4D spatio-temporal reasoning. Different from existing works that impose strong 3D geometric priors, RAYNOVA constructs an isotropic spatio-temporal representation across views, frames, and scales based on relative Plucker-ray positional encoding, enabling robust generalization to diverse camera setups and ego motions. We further introduce a recurrent training paradigm to alleviate distribution drift in long-horizon video generation. RAYNOVA achieves state-of-the-art multi-view video generation results on nuScenes, while offering higher throughput and strong controllability under diverse input conditions, generalizing to novel views and camera configurations without explicit 3D scene representation. Our code will be released at https://raynova-ai.github.io/.

Xiaofei Wu, Yi Zhang, Yumeng Liu, Yuexin Ma, Yujiao Shi, Xuming He

Generating human grasping poses that accurately reflect both object geometry and user-specified interaction semantics is essential for natural hand-object interactions in AR/VR and embodied AI. However, existing semantic grasping approaches struggle with the large modality gap between 3D object representations and textual instructions, and often lack explicit spatial or semantic constraints, leading to physically invalid or semantically inconsistent grasps. In this work, we present AffordGrasp, a diffusion-based framework that produces physically stable and semantically faithful human grasps with high precision. We first introduce a scalable annotation pipeline that automatically enriches hand-object interaction datasets with fine-grained structured language labels capturing interaction intent. Building upon these annotations, AffordGrasp integrates an affordance-aware latent representation of hand poses with a dual-conditioning diffusion process, enabling the model to jointly reason over object geometry, spatial affordances, and instruction semantics. A distribution adjustment module further enforces physical contact consistency and semantic alignment. We evaluate AffordGrasp across four instruction-augmented benchmarks derived from HO-3D, OakInk, GRAB, and AffordPose, and observe substantial improvements over state-of-the-art methods in grasp quality, semantic accuracy, and diversity.

Rishubh Parihar, Or Patashnik, Daniil Ostashev, Venkatesh Babu Radhakrishnan, Daniel Cohen-Or, Kuan-Chieh Jackson Wang

Instruction-based image editing offers a powerful and intuitive way to manipulate images through natural language. Yet, relying solely on text instructions limits fine-grained control over the extent of edits. We introduce Kontinuous Kontext, an instruction-driven editing model that provides a new dimension of control over edit strength, enabling users to adjust edits gradually from no change to a fully realized result in a smooth and continuous manner. Kontinuous Kontext extends a state-of-the-art image editing model to accept an additional input, a scalar edit strength which is then paired with the edit instruction, enabling explicit control over the extent of the edit. To inject this scalar information, we train a lightweight projector network that maps the input scalar and the edit instruction to coefficients in the model's modulation space. For training our model, we synthesize a diverse dataset of image-edit-instruction-strength quadruplets using existing generative models, followed by a filtering stage to ensure quality and consistency. Kontinuous Kontext provides a unified approach for fine-grained control over edit strength for instruction driven editing from subtle to strong across diverse operations such as stylization, attribute, material, background, and shape changes, without requiring attribute-specific training.

Hanyu Shi, Hong Tao, Guoheng Huang, Jianbin Jiang, Xuhang Chen, Chi-Man Pun, Shanhu Wang, Pan Pan

Unsupervised Concept Extraction aims to extract concepts from a single image, yet existing methods suffer from the inability to extract composable intrinsic concepts. To address this, this paper introduces a new task called Compositional and Interpretable Intrinsic Concept Extraction (CI-ICE). The CI-ICE task aims to leverage diffusion-based text-to-image models to extract composable object-level and attribute-level concepts from a single image, such that the original concept can be reconstructed through the combination of these concepts. To achieve this goal, we propose a method called HyperExpress, which addresses the CI-ICE task through two core aspects. Specifically, first, we propose a concept learning approach that leverages the inherent hierarchical modeling capability of hyperbolic space to achieve accurate concept disentanglement while preserving the hierarchical structure and relational dependencies among concepts; second, we introduce a concept-wise optimization method that maps the concept embedding space to maintain complex inter-concept relationships while ensuring concept composability. Our method demonstrates outstanding performance in extracting compositionally interpretable intrinsic concepts from a single image.

Bingxuan Li, Yiming Cui, Yicheng He, Yiwei Wang, Shu Zhang, Longyin Wen, Yulei Niu

Sound effects build an essential layer of multimodal storytelling, shaping the emotional atmosphere and the narrative semantics of videos. Despite recent advancement in video-text-to-audio (VT2A), the current formulation faces three key limitations: (1) an imbalance between visual and textual conditioning that leads to visual dominance; (2) the absence of a concrete definition for fine-grained controllable generation; (3) weak instruction understanding and following, as existing datasets rely on brief categorical tags. To address these limitations, we introduce EchoFoley (Event-Centric Hierarchical cOntrol Foley), a new task designed for video-grounded sound generation with both event-level local control and hierarchical semantic control. Our symbolic representation for sounding events specifies when, what, and how each sound is produced within a video or instruction, enabling fine-grained controls like sound generation, insertion, and editing. To support this task, we construct EchoFoley-6k, a large-scale, expert-curated benchmark containing over 6,000 video-instruction-annotation triplets and 42,000 fine-grained sounding event annotations.Building upon this foundation, we propose EchoVidia, a sounding-event-centric agentic generation framework with slow-fast thinking strategy. Experiments show that EchoVidia surpasses recent VT2A models by 40.7% in controllability and 12.5% in perceptual quality.

Jiaqi Tan, Xu Zheng, Yang Liu

Multimodal semantic segmentation (MMSS) faces significant challenges in real-world applications due to incomplete, degraded, or missing sensor data. To address this, we propose RobustSeg, an efficient teacher-student framework that enhances model robustness under missing-modality conditions while maintaining strong performance in full-modality scenarios. RobustSeg adopts a feedback-based self-distillation paradigm consisting of two complementary stages. Firstly, we introduce Hybrid Prototype Distillation (HPD), which enables more reliable knowledge transfer of both cross-modal and modality-specific aspects. Concretely, combined with dominant-modality selection, HPD performs cross-modal semantic distillation with high-level semantic prototypes to reduce modality bias. Meanwhile, HPD conducts intra-class feature variation distillation for modality-specific structural details. Secondly, to enable the teacher model to gradually produce more balanced and robust modality representations, we make the student model provide feedback from the non-dominant modality to the teacher, benefiting the entire distillation process. Experiments on three datasets demonstrate that our method achieves state-of-the-art robustness (e.g., +2.40% missing-modality performance on DeLiVER) while causing almost no degradation in full-modality performance (only -0.1% mIoU). Moreover, evaluations using different backbones (AnySeg and CMNeXt) further validate the generalization ability of RobustSeg.

Moyang Li, Zihan Zhu, Marc Pollefeys, Daniel Barath

We present a robust, real-time RGB SLAM system that handles dynamic environments by leveraging differentiable Uncertainty-aware Bundle Adjustment. Traditional SLAM methods typically assume static scenes, leading to tracking failures in the presence of motion. Recent dynamic SLAM approaches attempt to address this challenge using predefined dynamic priors or uncertainty-aware mapping, but they remain limited when confronted with unknown dynamic objects or highly cluttered scenes where geometric mapping becomes unreliable. In contrast, our method estimates per-pixel uncertainty by exploiting multi-view visual feature inconsistency, enabling robust tracking and reconstruction even in real-world environments. The proposed system achieves state-of-the-art camera poses and scene geometry in cluttered dynamic scenarios while running in real time at around 10 FPS. Code and datasets are available at https://github.com/MoyangLi00/DROID-W.git.

Guanjie Wang, Chen Chen

The recent success of artificial intelligence motivates many non-professional users to train their own models. Those users often resort to cloud training services, seeking to obtain a sufficiently accurate model at a modest cost, for which properly setting up the learning rate and batch size is crucial. While various Hyper-parameter Optimization (HPO) methods have been proposed in that regard, they largely act based on heavy-weight validation signals, being inefficient in the overall cost. We find that the model training process can be viewed as a two-dimensional voting process---with gradients for different iterations and from different samples; moreover, to attain cost-efficient training is to ensure that the gradient redundancy is within a proper range which is similar across diverse models. Based on that insight, we further introduce GR-Gauge, a general method that gauges the gradient redundancy to instruct HPO decisions like configuration searching and trial termination. Extensive experiments demonstrate that GR-Gauge can help attain near-optimal accuracy in much less time than existing methods.

Haoxiang Sun, Tao Wang, Chenwei Tang, Li Yuan, Jiancheng Lv

Following the success of Group Relative Policy Optimization (GRPO) in foundation LLMs, an increasing number of works have sought to adapt GRPO to Visual Large Language Models (VLLMs) for visual perception tasks (e.g., detection and segmentation). However, much of this line of research rests on a long-standing yet unexamined assumption: training paradigms developed for language reasoning can be transferred seamlessly to visual perception. Our experiments show that this assumption is not valid, revealing intrinsic differences between reasoning-oriented and perception-oriented settings. Using reasoning segmentation as a representative case, we surface two overlooked factors: (i) the need for a broader output space, and (ii) the importance of fine-grained, stable rewards. Building on these observations, we propose Dr.Seg, a simple, plug-and-play GRPO-based framework consisting of a Look-to-Confirm mechanism and a Distribution-Ranked Reward module, requiring no architectural modifications and integrating seamlessly with existing GRPO-based VLLMs. Extensive experiments demonstrate that Dr.Seg improves performance in complex visual scenarios while maintaining strong generalization. Code, data, and models will be publicly released upon acceptance.

Ziwei Xiang, Fanhu Zeng, Hongjian Fang, Rui-Qi Wang, Renxing Chen, Yanan Zhu, Yi Chen, Peipei Yang, Xu-Yao Zhang

Large Vision Language Models (LVLMs) have achieved remarkable success in a wide range of downstream tasks that require multimodal interaction, but their powerful capabilities come with substantial computational and memory overhead, which hinders practical deployment. Among numerous acceleration techniques, post-training quantization is a popular and effective strategy for reducing memory cost and accelerating inference. However, existing LVLM quantization methods typically measure token sensitivity at the modality level, which fails to capture the rich and complex cross-token interactions and falls short in quantitatively measuring the quantization error at the token level. As tokens interact within the model, the distinction between modalities gradually diminishes, suggesting the need for fine-grained calibration. Inspired by axiomatic attribution in mechanistic interpretability, we introduce a fine-grained quantization strategy based on Quantization-aware Integrated Gradients (QIG), which leverages integrated gradients to quantitatively evaluate token sensitivity and push the granularity from modality level to token level, reflecting both inter-modality and intra-modality dynamics. Extensive experiments on multiple LVLMs under both W4A8 and W3A16 settings show that our method consistently improves accuracy across models and benchmarks with negligible latency overhead. For example, under 3-bit weight-only quantization, our method improves the average accuracy of LLaVA-onevision-7B by 1.60%, reducing the gap to its full-precision counterpart to only 1.33%. The code is available at https://github.com/ucas-xiang/QIG.

An-Chieh Cheng, Yang Fu, Yatai Ji, Ligeng Zhu, Guanqi Zhan, Zhuoyang Zhang, Zhaojing Yang, Song Han, Yao Lu, Pavlo Molchanov 等

We present GR3D, a spatial vision language model equipped with three complementary grounding capabilities--explicit 2D grounding, implicit 2D grounding, and monocular 3D grounding--within a single framework. GR3D introduces an implicit grounding mechanism that identifies entity mentions during generation and inserts the corresponding region tokens into the text stream, allowing the model to reference visual evidence on the fly when producing spatial chain-of-thought responses. In parallel, a region-prompted monocular 3D grounding design predicts 3D bounding boxes in the camera view from grounded region queries, supported by intrinsic-aware normalization and dense geometric supervision. Together, these grounding capabilities enable GR3D to decompose complex spatial understanding problems into grounded 2D perception followed by 3D inference. GR3D achieves consistent improvements across grounded and non-grounded spatial benchmarks, demonstrating grounding as an effective inductive bias for strengthening spatial understanding in VLMs. These grounding capabilities collectively enhance general spatial understanding beyond the grounding task itself.