cs.CV（2026-10-15）

📊 共 62 篇论文 | 🔗 11 篇有代码

🎯 兴趣领域导航

支柱九：具身大模型 (Embodied Foundation Models) (26 🔗5) 支柱二：RL算法与架构 (RL & Architecture) (17 🔗2) 支柱三：空间感知与语义 (Perception & Semantics) (8 🔗3) 支柱七：动作重定向 (Motion Retargeting) (6) 支柱一：机器人控制 (Robot Control) (3 🔗1) 支柱八：物理动画 (Physics-based Animation) (2)

🔬 支柱九：具身大模型 (Embodied Foundation Models) (26 篇)

#题目一句话要点标签🔗⭐
1 Do VLAs Really Follow Instructions? Diagnosing How-Level Instruction Following in Vision-Language-Action Models 提出改进方法以解决视觉语言行动模型的指令遵循问题 vision-language-action VLA instruction following
2 VLA-$Δ$: Learning Physical State Transition for Vision-Language-Action Models 提出VLA-$Δ$以解决视觉-语言-动作模型的状态转移问题 vision-language-action VLA ✅
3 FuseAdapt: Parameter-Efficient Multimodal Fusion for Semantic Segmentation with Missing Modalities 提出FuseAdapt以解决多模态语义分割中的缺失模态问题 foundation model multimodal ✅
4 Pathology foundation models for training-free case retrieval in salivary gland tumors 提出基于病理基础模型的无训练病例检索方法以解决唾液腺肿瘤诊断问题 foundation model
5 From Adults to Children: Benchmarking Foundation Models for Chest X-Ray Classification under Domain Shift 基于基础模型的胸部X光分类研究解决成人到儿童的领域转移问题 foundation model
6 Beyond Generation: Using Multimodal LLMs as Detectors of LLM-Generated Deepfake Facial Images 利用多模态LLM检测深度伪造人脸图像 multimodal
7 FusionAudit: Pathway-Conditioned Robustness Auditing for Native Multimodal Models 提出FusionAudit以解决多模态模型的鲁棒性审计问题 multimodal
8 Can Visual Generation Expand What Unified Multimodal Models Know? 提出知识转移探测以扩展统一多模态模型的知识 multimodal
9 Loud and Clear: Making Visual Evidence Count in Multimodal Document RAG 提出Loud and Clear方法以提升多模态文档RAG的视觉证据利用率 multimodal
10 TRACER: A Multimodal Harness for Patient Care in Hospitals 提出TRACER以解决医院患者行为理解问题 multimodal
11 Grounding Generative Verification in 3D Content 提出3DVerifierBench以解决3D内容生成中的验证问题 large language model multimodal
12 MedPAO-VL: Interpretable and Protocol-Compliant Agent for Structured Radiology Report Generation 提出MedPAO-VL以解决结构化放射学报告生成中的可靠性问题 large language model multimodal
13 VisualTrust: Video Reasoning under On-Screen Claim Framing 提出VisualTrust以解决视频推理中的文本可信度问题 large language model multimodal
14 MJev-Spatial: System One Spatial Intelligence via Native Attention Matching 提出MJev-Spatial以解决空间智能决策问题 multimodal
15 WaveLLM: Wavelet-Domain LLM Entropy Modeling for Lossless Image Compression 提出WaveLLM以解决无损图像压缩中的熵建模问题 large language model
16 Memory as Weights: Internalizing Long-Term History for Streaming Videos 提出MaLoW框架以解决流媒体视频中的长期历史信息整合问题 vision-language-action
17 Visual Sanad: Auditing Vision-Language Models for False Acceptance of Qur'anic Citations 提出Visual Sanad以解决视觉语言模型的错误接受问题 multimodal
18 Towards Reconstructing Geographically Diverse Architecture with 3D Foundation Models 提出ArchWorld基准以解决3D基础模型地理差异问题 foundation model
19 Star Light, Star Bright: Can Multimodal Models Parse an Astronomy Visual Question Answering Dataset Right? 构建天文学视觉问答数据集以评估多模态模型的表现 multimodal ✅
20 Beyond Generation: Using Multimodal LLMs as Detectors of LLM-Generated Deepfake Facial Images 利用多模态大语言模型检测深度伪造人脸图像 multimodal
21 Thinking from Above: Efficient Spatial Reasoning with Bird's-Eye-View Visual Tokens 提出Cambrian-BEV以解决视频多模态大语言模型空间推理问题 large language model multimodal
22 Event-STVG: Spatio-Temporal Video Grounding for Long-Form Event Reasoning 提出Event-STVG以解决长视频事件推理问题 large language model multimodal ✅
23 TriSAR-3D and PaintFusion: A Novel Trimodal LiDAR-RGB-Thermal Dataset and Method for 3D Person Detection in Search and Rescue 提出TriSAR-3D与PaintFusion以解决救援场景中的3D人检测问题 multimodal ✅
24 Stuck on Shuffle: On Attribute-Object Binding in Vision-Language Models 提出SABER基准以解决多模态模型中的属性-对象绑定问题 multimodal
25 JAVIS: Joint Audio-Visual Interaction and Streaming for Real-Time Conversational Agents 提出JAVIS以解决实时对话系统中的多模态交互问题 multimodal
26 WaveLLM: Wavelet-Domain LLM Entropy Modeling for Lossless Image Compression 提出WaveLLM以解决无损图像压缩中的熵建模问题 large language model

🔬 支柱二：RL算法与架构 (RL & Architecture) (17 篇)

#题目一句话要点标签🔗⭐
27 Rethinking Classifier-Free Guidance for TI2I Flow Matching 提出改进的无分类器引导以优化TI2I流匹配 flow matching classifier-free guidance instruction following
28 Motion- and Gravity-Aware Feedforward 3D Reconstruction 提出运动与重力感知框架以提升动态场景3D重建精度 representation learning privileged information 3D reconstruction
29 GeoState: Learning View-Decoupled Geometry States for 3D World Modeling 提出GeoState以解决3D世界建模中的视角依赖问题 world model world models foundation model ✅
30 MMSearch-OPD: Easy Made Hard, Hard Made Learnable for Long-Horizon Multimodal Search Agents 提出MMSearch-OPD以解决长时间多模态搜索代理的训练挑战 reinforcement learning distillation multimodal
31 Velox: Spatial Memory Makes World Models Efficient 提出Velox以解决视频世界模型生成效率问题 world model world models
32 Skill-Decoupled On-Policy Expert Distillation for Unified Vision-Language Models 提出SCOPED以解决视觉语言模型中的能力干扰问题 reinforcement learning distillation multimodal ✅
33 SCCL-DT: Semantic-Conflict-Aware Contrastive Learning with Dual-Rule Teacher Distillation for Zero-Shot Chest X-Ray Diagnosis 提出SCCL-DT以解决胸部X光零-shot诊断中的语义冲突问题 contrastive learning distillation
34 Perception or Reasoning? Diagnosing and Tackling Reinforcement Learning Bottlenecks in VLMs 提出PARA和GRAD以解决视觉语言模型中的强化学习瓶颈问题 reinforcement learning distillation multimodal
35 ControlFlow3D: Distilling Multi-View Knowledge into Latent Flow Matching for Point Cloud Upsampling 提出ControlFlow3D以解决点云上采样问题 flow matching Mamba foundation model
36 FocusOPD: Structure-Guided On-Policy Distillation for Diffusion Models 提出FocusOPD以解决扩散模型蒸馏中的特征对齐问题 teacher-student distillation
37 RoPE-Flow: Monocular Vision-based Robot Pose Estimation via 2D Graph-Conditioned 3D Flow Matching 提出RoPE-Flow以解决单目视觉机器人姿态估计问题 flow matching
38 CoDeFont: One-Shot Font Generation via Multi-Scale Factor-Specific Representation Learning 提出CoDeFont以解决单次字体生成中的风格捕捉难题 representation learning
39 Can World and Action Modeling Benefit Each Other under Unified Causal Training? 提出OneWM框架以促进世界与动作建模的相互提升 policy learning world model world models
40 Learning to Think and Explore in 3D World 提出ACT3D以解决3D空间中的视觉语言模型探索问题 reinforcement learning first-person view
41 OmniXRef: A Unified Framework for Multimodal Multi-Reference Video Editing 提出OmniXRef框架以解决多模态视频编辑问题 curriculum learning multimodal
42 EgoP$^{2}$C: Internalizing Plan-Perceive-Compose for Spatial Reasoning in Egocentric Video 提出EgoP$^{2}$C以解决自我中心视频中的空间推理问题 distillation egocentric
43 Is a Linear Probe Evidence of a Linear Representation? 提出新诊断方法以区分线性表示与曲面结构 world model world models

🔬 支柱三：空间感知与语义 (Perception & Semantics) (8 篇)

#题目一句话要点标签🔗⭐
44 Anchor3DGS: Feed-Forward 3D Gaussian Splatting with Compact Anchor-Based Representation 提出Anchor3DGS以解决3D重建中的高冗余与覆盖不足问题 3D gaussian splatting 3DGS 3D reconstruction
45 Hear, Localize, and Reason: Spatially Aware Scene Understanding for Audio-Visual LLMs 提出HLR-AVSceneQA和HLR-LLM以解决音频视觉模型空间理解问题 scene understanding egocentric large language model
46 GeoPano: Towards Geometrically Accurate 3D Reconstruction from a Single Panorama 提出GeoPano以解决全景图像几何重建不一致问题 3D reconstruction geometric consistency
47 Edit4D: Editable 4D Scene Generation with Textual Control 提出Edit4D以解决可编辑4D场景生成问题 3D gaussian splatting gaussian splatting splatting ✅
48 BridgeAfford: Bridging Query-Target Gap for Generalizable 3D Affordance Grounding 提出BridgeAfford以解决3D物体功能区域定位的泛化问题 affordance
49 DK-VLM: Geometry as Position for Spatial Reasoning in Vision-Language Models 提出DK-VLM以解决视觉语言模型的空间推理问题 3D reconstruction ✅
50 VesperMUSE: Multi-Modal Fusion Vision Transformer for Image Semantic Segmentation Under Diverse Light Conditions 提出VesperMUSE以解决低光环境下的语义分割问题 scene understanding
51 KFS$^4$LAM: Adaptive KeyFrame Selection with Structure-Aware Mapping for Endoscopic Gaussian Splatting SLAM 提出KFS$^4$LAM以解决内窥镜SLAM中的关键帧选择问题 gaussian splatting splatting SplaTAM ✅

🔬 支柱七：动作重定向 (Motion Retargeting) (6 篇)

#题目一句话要点标签🔗⭐
52 FutureLift: Lifting Trajectory Proposals into Bidirectional Future Graphs for Motion Forecasting 提出FutureLift以解决自主驾驶中的运动预测精度问题 motion prediction multimodal
53 Dense4D: Pixel-Level 4D Human Correspondence 提出Dense4D以解决4D人类运动重建问题 human motion
54 ARC-SORT: Adaptive, Robust, Compensated Multi-Object Tracking with Interacting Multiple Models 提出ARC-SORT以解决体育赛事中的多目标跟踪问题 motion prediction
55 CLAD: Keypoint Matching Using Global Line Correspondences 提出CLAD以解决重复模式下关键点匹配问题 spatial relationship
56 GRASE: Geometry-native Real-to-sim Agentic Scene Modeling from Video 提出GRASE以解决真实场景建模中的一致性问题 spatial relationship
57 DLOSAC: Joint Spatial-Geometric Image Estimation via Delaunay-Guided Local Optimization RANSAC 提出DLOSAC以解决噪声匹配点的几何估计问题 geometric consistency

🔬 支柱一：机器人控制 (Robot Control) (3 篇)

#题目一句话要点标签🔗⭐
58 Perceive, Predict, and Plan: a Unified World Action Model for Egocentric Dexterous Manipulation 提出P³-WAM模型以解决自我中心灵巧操作中的感知与规划问题 manipulation dexterous manipulation world action model
59 RLDF: A Region-Level Deepfake Dataset and Benchmark for Partial-Face Manipulation Detection 提出RLDF数据集以解决局部面部篡改检测问题 manipulation ✅
60 An Agentic ENF Framework for Forensic Video Authentication Against Generative AI 提出基于ENF框架的视频认证方法以应对生成式AI挑战 manipulation

🔬 支柱八：物理动画 (Physics-based Animation) (2 篇)

#题目一句话要点标签🔗⭐
61 Dusting the Phantom: Gaussian Allocation for Fading Dynamic Content 提出高斯分配策略以解决动态内容重建问题 spatiotemporal
62 Dusting the Phantom: Gaussian Allocation for Fading Dynamic Content 提出高斯分配策略以解决动态视频重建中的快速运动物体消失问题 spatiotemporal

⬅️ 返回 cs.CV 首页 · 🏠 返回主页