CVPR 2026 论文列表

Understanding the fine-grained articulation of human hands is critical in high-stakes settings such as robot-assisted surgery, chip manufacturing, and AR/VR-based human-AI interaction. Despite achieving near-human performance on general vision-language benchmarks, current vision-language models (VLMs) struggle with fine-grained spatial reasoning, especially in interpreting complex, articulated hand poses. We introduce HandVQA, a large-scale diagnostic benchmark designed to evaluate VLMs' understanding of detailed hand anatomy through visual question answering. Built upon high-quality 3D hand datasets (FreiHAND, InterHand2.6M, FPHA), our benchmark includes over 1.6M controlled multiple-choice questions that probe spatial relationships between hand joints, such as angles, distances, and relative positions. We evaluate several state-of-the-art VLMs (LLaVA, DeepSeek and Qwen-VL) in both base and fine-tuned settings, using lightweight fine-tuning via LoRA. Our findings reveal systematic limitations in current models, including hallucinated finger parts, incorrect geometric interpretations, and poor generalization. HandVQA not only exposes these critical reasoning gaps but provides a validated path to improvement. We demonstrate that the 3D-grounded spatial knowledge learned from our benchmark transfers in a zero-shot setting, significantly improving accuracy of model on novel downstream tasks like hand gesture recognition (+10.33%) and hand-object interaction (+2.63%).

Recent vision-language models (VLMs) achieve remarkable reasoning through reinforcement learning (RL), which provides a feasible solution for realizing continuous self-evolving large vision-language models (LVLMs) in the era of experience. However, RL for VLMs requires abundant high-quality multimodal data--especially challenging in specialized domains like chemistry, earth sciences, and multimodal mathematics. Existing strategies such as synthetic data and self-rewarding mechanisms suffer from limited distributions and alignment difficulties, ultimately causing reward hacking: models exploit high-reward patterns, collapsing policy entropy and destabilizing training. We propose DoGe (Decouple to Generalize), a dual-decoupling framework that guides models to first learn from context rather than problem solving by refocusing on the problem context scenarios overlooked by synthetic data methods. By decoupling learning process into dual components (Thinker and Solver), we reasonably quantify the reward signals of this process and propose a two-stage RL post-training approach from freely exploring context to practically solving tasks. Second, to increase the diversity of training data, DoGe constructs an evolving curriculum learning pipeline: an expanded native domain knowledge corpus and an iteratively evolving seed problems pool. Experiments show that our method consistently outperforms the baseline across various benchmarks, providing a scalable pathway for realizing self-evolving LVLMs.

Recent advancements in reasoning language models have fueled growing interest in extending such capabilities to multimodal domains. However, despite notable progress in visual and video reasoning, the lack of transparent and reproducible data curation and training pipelines remains a major barrier to scalable research. In this work, we introduce OpenMMReasoner, a fully transparent two-stage recipe for multimodal reasoning spanning supervised fine-tuning (SFT) and reinforcement learning (RL). In the SFT stage, we construct an 874k-sample cold-start dataset with rigorous step-by-step validation, providing a strong foundation for reasoning capabilities. The subsequent RL stage leverages a 74k-sample dataset across diverse domains to further sharpen and stabilize these abilities, resulting in a more robust and efficient learning process. Extensive evaluations demonstrate that our training recipe not only surpasses strong baselines but also highlights the critical role of data quality and training design in shaping multimodal reasoning performance. Notably, our method achieves a 9.5% improvement over the Qwen2.5-VL-7B-Instruct baseline across nine multimodal reasoning benchmarks, establishing a solid empirical foundation for future large-scale multimodal reasoning research.

Reading text from images or scanned documents via OCR models has been a longstanding focus of researchers. Intuitively, text reading is perceived as a straightforward perceptual task, and existing work primarily focuses on constructing enriched data engineering to enhance SFT capabilities. In this work, we observe that even advanced OCR models exhibit significantly higher entropy in formatted text (e.g., formula, table) compared to plain text, often by an order of magnitude. These statistical patterns reveal that advanced OCR models struggle with high output uncertainty when dealing with format sensitive document, suggesting that reasoning over diverse reading pathways may improve OCR performance. To address this, we propose Format Decoupled Reinforcement Learning (FD-RL) , which leverages high-entropy patterns for targeted optimization. Our approach employs entropy-based data filtration strategy to identify format-intensive instances, and adopt format decoupled rewards tailored to different format types, enabling format-level validation rather than token-level memorization. FD-RL achieves an average score of 90.41 on OmniDocBench, demonstrating highly competitive performance among end-to-end models on this highly popular benchmark. More importantly, we conduct comprehensive ablation studies over data, training, filtering, and rewarding strategies, thoroughly validating their effectiveness.

Diffusion large language models (dLLMs) are emerging as promising alternatives to autoregressive (AR) LLMs. Recently, this paradigm has been extended to multimodal tasks, leading to the development of diffusion multimodal large language models (dMLLMs). These models are expected to retain the reasoning capabilities of LLMs while enabling faster inference through parallel generation. However, when combined with Chain-of-Thought (CoT) reasoning, dMLLMs exhibit two critical issues.First, we observe that dMLLMs often generate the final answer token at a very early timestep. This trend indicates that the model determines the answer before sufficient reasoning, leading to degraded reasoning performance. Second, during the initial timesteps, dMLLMs show minimal dependency on visual prompts, exhibiting a fundamentally different pattern of visual information utilization compared to AR vision-language models. In summary, these findings indicate that dMLLMs tend to generate premature final answers without sufficiently grounding on visual inputs.To address these limitations, we propose Position & Step Penalty (PSP) and Visual Reasoning Guidance (VRG). PSP penalizes tokens in later positions during early timesteps, delaying premature answer generation and encouraging progressive reasoning across timesteps. VRG, inspired by the classifier-free guidance, amplifies visual grounding signals to enhance the model's alignment with visual evidence.Extensive experiments across various dMLLMs demonstrate that our method achieves up to 7.5% higher accuracy while delivering more than 3x speedup compared to reasoning with four times more diffusion steps. Our code will be released after publication.

This paper introduces UniMERNet, a high-accuracy, computation-efficient algorithm for Mathematical Expression Recognition (MER) across diverse real-world scenarios. To facilitate UniMERNet's training, we constructed UniMER-1M, a million-scale dataset whose unprecedented diversity endows the model with robust generalization ability. Through in-depth analysis, we discover a distinctive raster-scan pattern (left-to-right, top-to-bottom) in the attention distribution of Transformer models for MER tasks, which closely aligns with human reading habits. Based on this key finding, we design an innovative **Raster-Scan Attention** mechanism that employs a "horizontal-first, vertical-second" sequential attention computation strategy. This approach not only successfully reduces computational complexity from \mathcal O (NH^2W^2D) to \mathcal O (NHWD(H + W)) , but also enables the model to capture long-range dependencies more efficiently, achieving recognition performance comparable to global attention. Leveraging both UniMER-1M and our innovative attention mechanism, UniMERNet achieves state-of-the-art performance across four real-world scenarios while significantly reducing computational resources compared to global attention: over **1.2x** memory savings during training, approximately **10x** memory reduction during inference, and **5x** speed with slightly improved accuracy. All resources will be publicly released to advance MER research further.

Vision-language models (VLMs) have achieved remarkable success across diverse tasks. However, concerns about their trustworthiness persist, particularly regarding tendencies to lean more on textual cues than visual evidence and the risk of producing ungrounded or fabricated responses. To address these issues, we propose Saliency-R1, a framework for improving the interpretability and faithfulness of VLMs reasoning. Specifically, we introduce a novel saliency map technique that efficiently highlights critical image regions contributing to generated tokens without additional computational overhead. This can further be extended to trace how visual information flows through the reasoning process to the final answers, revealing the alignment between the thinking process and the visual context. We use the overlap between the saliency maps and human-annotated bounding boxes as the reward function, and apply Group Relative Policy Optimization (GRPO) to align the salient parts and critical regions, encouraging models to focus on relevant areas when conduct reasoning. Experiments show Saliency-R1 improves reasoning faithfulness, interpretability, and overall task performance.

Tagged MRI enables tracking internal tissue motion non-invasively. It encodes motion by modulating anatomy with periodic tags, which deforms along with tissue. However, the entanglement between anatomy, tags and motion poses significant challenges on post processing. Existence of tags and imaging blur hinders downstream tasks such as segmenting anatomy. Tag fading, due to T1-relaxation, disrupts brightness constancy assumption for motion tracking. For decades, these challenges are handled in isolation and sub-optimally. In contrast, we introduce a blind and nonlinear inverse framework for tagged MRI that, for the first time, unifies these tasks: anatomical image recovery, high-resolution cine image synthesis, and motion estimation. At its core, the synergy of MR physics and generative priors enables us to blindly estimate the unknown forward imaging models, high-resolution underlying anatomy, while simultaneously tracking 3D diffeomorphic Lagrangian motion over time. Experiments on tagged brain MRI demonstrate that our approach yields high-resolution anatomy images, cine images, and more accurate motion than specialized methods.

Vision-Language Models (VLMs) have been increasingly applied in real-world scenarios due to their outstanding understanding and reasoning capabilities. Although VLMs have already demonstrated impressive capabilities in common visual question answering and logical reasoning, they still lack the ability to make reasonable decisions in complex real-world environments. We define this ability as spatial logical reasoning, which not only requires understanding the spatial relationships among objects in complex scenes, but also the logical dependencies between steps in multi-step tasks. To bridge this gap, we introduce Spatial Logical Question Answering (SpatiaLQA), a benchmark designed to evaluate the spatial logical reasoning capabilities of VLMs. SpatiaLQA consists of 9,605 question answer pairs derived from 241 real-world indoor scenes. We conduct extensive experiments on 41 mainstream VLMs, and the results show that even the most advanced models still struggle with spatial logical reasoning. To address this issue, we propose a method called recursive scene graph assisted reasoning, which leverages visual foundation models to progressively decompose complex scenes into task-relevant scene graphs, thereby enhancing the spatial logical reasoning ability of VLMs, outperforming all previous methods. We will release our code and dataset soon.

Masked Discrete Diffusion Models (MDMs) have achieved strong performance across a wide range of multimodal tasks, including image understanding, generation, and editing. However, their inference speed remains suboptimal due to the need to repeatedly process redundant masked tokens at every sampling step. In this work, we propose Sparse-LaViDa, a novel modeling framework that dynamically truncates unnecessary masked tokens at each inference step to accelerate MDM sampling. To preserve generation quality, we introduce specialized register tokens that serve as sparse representations for the truncated tokens. Furthermore, to ensure consistency between training and inference, we design a specialized attention mask that faithfully matches the truncated sampling procedure during training. Built upon the state-of-the-art unified MDM LaViDa-O, Sparse-LaViDa achieves up to a 2xspeedup across diverse tasks including text-to-image generation, image editing, and mathematical reasoning, while maintaining generation quality.

Most referring object detection (ROD) models, especially the modern grounding detectors, are designed for data-rich conditions, yet many practical deployments, such as robotics, augmented reality, and other specialized domains, would face severe label scarcity. In such regimes, end-to-end grounding detectors need to learn spatial and semantic structure from scratch, wasting precious samples. We ask a simple question: Can explicit reasoning priors help models learn more efficiently when data is scarce? To explore this, we first introduce a Data-efficient Referring Object Detection (De-ROD) task, which is a benchmark protocol for measuring ROD performance in low-data and few-shot settings. We then propose the HeROD (Heuristic-inspired ROD), a lightweight, model-agnostic framework that injects explicit, heuristic-inspired spatial and semantic reasoning priors, which are interpretable signals derived based on the referring phrase, into 3 stages of a modern DETR-style pipeline: proposal ranking, prediction fusion, and Hungarian matching. By biasing both training and inference toward plausible candidates, these priors promise to improve label efficiency and convergence performance. On RefCOCO, RefCOCO+, and RefCOCOg, HeROD consistently outperforms strong grounding baselines in scarce-label regimes. More broadly, our results suggest that integrating simple, interpretable reasoning priors provides a practical and extensible path toward better data-efficient vision-language understanding. Code at: https://github.com/xuzhang1199/HeROD.

Large multimodal models (LMMs) have achieved impressive performance on multimodal reasoning, becoming crucial technology for the advancement of intelligent question-answering systems. In real-world educational scenarios, effective teaching extends far beyond providing answers. Experienced teachers analyze students' incorrect answers to trace underlying errors and provide corrective feedback, termed educational diagnostic reasoning, a capability that remains under-explored in existing LMMs. To bridge this research gap, we introduce Edudiag benchmark, requiring LMMs to reconstruct erroneous reasoning chains from incorrect answers and generate corrective feedback. Through an AI-assisted annotation pipeline with rigorous human verification, we create 8K erroneous reasoning chains and corresponding feedback, spanning three representative educational domains: commonsense, science, and mathematics. Extensive evaluation across 28 leading LMMs highlights Edudiag as a challenging testbed, where even leading proprietary LMMs struggle on it and supervised fine-tuning (SFT) on open-source LMMs achieves marginal performance gains. Moreover, we conduct analysis experiments and identify three critical insights for educational diagnostic reasoning: (i) Effective error tracing remains the primary bottleneck, while SFT models still fail to reversely identify errors that commonly occur. (ii) Group relative policy optimization (GRPO) mitigates this bottleneck and boosts performance. (iii) LMMs optimized with GRPO can generate plausible yet challenging distractors for multiple-choice questions based on their self-constructed erroneous reasoning chains. We believe Edudiag provides a new direction for evaluating the advanced LMMs.

Fine-grained spatiotemporal reasoning on surgical videos is critical, yet the capabilities of Multi-modal Large Language Models (MLLMs) in this domain remain largely unexplored. To bridge this gap, we introduce SurgCoT, a unified benchmark for evaluating chain-of-thought (CoT) reasoning in MLLMs across 7 surgical specialties and 35 diverse procedures. SurgCoT assesses five core reasoning dimensions: Causal Action Ordering, Cue-Action Alignment, Affordance Mapping, Micro-Transition Localization, and Anomaly Onset Tracking, through a structured CoT framework with an intensive annotation protocol (Question-Option-Knowledge-Clue-Answer), where the Knowledge field provides essential background context and Clue provides definitive spatiotemporal evidence. Evaluation of 10 leading MLLMs shows: 1) commercial models outperform open-source and medical-specialized variants; 2) significant gaps exist in surgical CoT reasoning; 3) SurgCoT enables effective evaluation and enhances progressive spatiotemporal reasoning. SurgCoT provides a reproducible testbed to narrow the gap between MLLM capabilities and clinical reasoning demands. Code and data: https://github.com/CVI-SZU/SurgCoT.

Current Vision-Language-Action (VLA) models primarily focus on mapping 2D observations to actions but exhibit notable limitations in spatiotemporal perception and reasoning: 1) spatial representations often rely on additional sensors, introducing substantial computational overhead; 2) visual reasoning is typically limited to future-frame prediction, lacking alignment with the instruction-grounded scene and thus compromising spatiotemporal consistency. To address these challenges, we propose **ConsisVLA-4D**, a unified and efficient framework that enhances spatiotemporal consistency in 3D-Perception and 4D-Reasoning. Specifically, we design: **1) CV-Aligner**, which ensures **C**ross-**V**iew object semantic consistency via filtering instruction-relevant regions and aligning object identities across multiple viewpoints; **2) CO-Fuser**, which guarantees **C**ross-**O**bject spatial geometric consistency by eliminating spatial relation ambiguities between objects across views using compact latent representations. Building upon these, we introduce **3) CS-Thinker** to achieve **C**ross-**S**cene spatiotemporal consistency as actions unfold. It learns implicit knowledge of local dynamics from object-semantic tokens of CV-Aligner and global depth from geometric tokens of CO-Fuser, thereby enhancing efficient visual reasoning under scene variations. Extensive experiments demonstrate that, benefiting from its efficient spatiotemporal consistency design, ConsisVLA-4D achieves **21.6%** and **41.5%** performance improvements, along with **2.3x** and **2.4x** inference speedups compared to OpenVLA on the LIBERO benchmark and real-world platforms, respectively.

Handwritten mathematical expression recognition is hindered by a fundamental misalignment between the dual representations of LaTeX formulas: the symbolic text and the rendered visual image. This discrepancy means that textually distinct LaTeX sequences can produce visually identical outputs, while minor textual errors can cause catastrophic rendering failures. As a result, text-level reward mechanisms cannot perfectly assess the quality of model predictions, failing to effectively guide the model towards optimal performance during training. To overcome this limitation, we introduce the Image Matching Score (IMS), a lightweight yet effective reward based on the structural edit distance of column-wise image projections, which robustly quantifies the visual fidelity between rendered formulas. Leveraging IMS, we then propose Image-Matching driven Policy Optimization (IMPO), a training framework built upon Group Relative Policy Optimization (GRPO). This approach facilitates stable policy learning directly from our sequence-level visual reward, notably without the need for a separate value function network. Extensive experiments demonstrate that IMPO yields consistent performance gains across various backbone models on the challenging CROHME, HME100K, and M2E datasets. Our framework establishes new state-of-the-art results, improving the Expression Recognition Rate by an average of 1.1% and up to 1.37% over strong prior methods.

Vision-Language Models (VLMs) have achieved remarkable progress across tasks such as visual question answering and image captioning. Yet, the extent to which these models perform visual reasoning as opposed to relying on linguistic priors remains unclear. To address this, we introduce VisRes Bench, a benchmark designed to study visual reasoning in naturalistic settings without contextual language supervision. Analyzing model behavior across three levels of complexity, we uncover clear limitations in perceptual and relational visual reasoning capacities. VisRes isolates distinct reasoning abilities across its levels. Level 1 probes perceptual completion and global image matching under perturbations such as blur, texture changes, occlusion, and rotation; Level 2 tests rule-based inference over a single attribute (e.g., color, count, orientation); and Level 3 targets compositional reasoning that requires integrating multiple visual attributes. Across more than 19,000 controlled task images, we find that state-of-the-art VLMs perform near random under subtle perceptual perturbations, revealing limited abstraction beyond pattern recognition. We conclude by discussing how VisRes provides a unified framework for advancing abstract visual reasoning in multimodal research (https://visres-bench.github.io).

While Large Multimodal Models (LMMs) have made significant progress, they remain largely text-centric, relying on language as their core reasoning modality. As a result, they are limited in their ability to handle reasoning tasks that are predominantly visual. Recent approaches have sought to address this by supervising intermediate visual steps with helper images, depth maps, or image crops. However, these strategies impose restrictive priors on what "useful" visual abstractions look like, add heavy annotation costs, and struggle to generalize across tasks. To address this critical limitation, we propose a task-agnostic mechanism that trains LMMs to discover and use visual reasoning tokens without explicit supervision. These tokens attend globally and re-encode the image in a task-adaptive way, enabling the model to extract relevant visual information without hand-crafted supervision. Our approach outperforms direct fine-tuning and achieves state-of-the-art results on a diverse range of vision-centric tasks -- including those where intermediate abstractions are hard to specify -- while also generalizing to multi-task instruction tuning.

Wide-baseline matching (WBM) requires integrating geometric understanding, viewpoint changes, fine-grained perception, and occlusion reasoning, making it a challenging testbed for spatial reasoning in multimodal large language models (MLLMs) deployed in physical environments. However, current MLLMs lack systematic evaluation and training frameworks for these capabilities. We introduce ReasonMatch-Bench, a benchmark stratified by viewpoint displacement and matching granularity across indoor, outdoor, and object-centric scenarios, and show that current MLLMs still struggle with fine-grained wide-baseline correspondence: on a difficult 90-sample subset, human annotators achieve 84.0 F1, while the best existing baseline reaches 37.2. To bridge this gap, we build a scalable data-generation pipeline that automatically extracts wide-baseline view pairs from large-scale video-3D corpora, including RGB-D videos and SfM reconstructions, yielding diverse and verifiable supervision. We further propose Dynamic Correspondence Reinforcement Learning (DCRL), which combines Image-Level Viewpoint Progression and Point-Level Correspondence Curriculum to improve WBM training through verifiable rewards without explicit CoT supervision. Extensive experiments show that DCRL substantially improves ReasonMatch-Bench and transfers to related spatial benchmarks, while maintaining general visual understanding performance with modest gains on several benchmarks.

Reinforcement Learning with Verifiable Rewards (RLVR) has become a key paradigm to improve the reasoning capabilities of Multimodal Large Language Models (MLLMs). However, prevalent group-based algorithms such as GRPO require multi-rollout sampling for each prompt. While more efficient single-rollout variants have recently been explored in text-only settings, we find that they suffer from severe instability in multimodal contexts, often leading to training collapse. To address this sample efficiency-stability trade-off, we introduce MSSR (Multimodal Stabilized Single-Rollout), a group-free RLVR framework that achieves both stable optimization and effective multimodal reasoning performance. MSSR achieves this via an entropy-based advantage-shaping mechanism that adaptively regularizes advantage magnitudes, preventing collapse and maintaining training stability. While such mechanisms have been used in group-based RLVR, we show that in the multimodal single-rollout setting they are not merely beneficial but essential for stability. In in-distribution evaluations, MSSR demonstrates superior rollout sample efficiency, achieving similar validation accuracy with half the training steps. When trained for the same number of steps, MSSR's performance surpasses the group-based baseline and shows consistent generalization improvements across five diverse reasoning-intensive benchmarks. Together, these results demonstrate that MSSR enables stable, sample-efficient, and effective RLVR for complex multimodal reasoning tasks.