cs.CV（2026-09-30）

📊 共 71 篇论文 | 🔗 19 篇有代码

🎯 兴趣领域导航

支柱二：RL算法与架构 (RL & Architecture) (29 🔗7) 支柱九：具身大模型 (Embodied Foundation Models) (21 🔗7) 支柱三：空间感知与语义 (Perception & Semantics) (9 🔗1) 支柱一：机器人控制 (Robot Control) (7 🔗3) 支柱八：物理动画 (Physics-based Animation) (3) 支柱六：视频提取与匹配 (Video Extraction) (1 🔗1) 支柱五：交互与反应 (Interaction & Reaction) (1)

🔬 支柱二：RL算法与架构 (RL & Architecture) (29 篇)

#题目一句话要点标签🔗⭐
1 MCD: Causal Distillation of Multimodal In-Context Learning in Large Vision-Language Models 提出多模态因果蒸馏方法以提升视觉语言模型的学习能力 distillation multimodal
2 Is Better Teacher Supervision Enough? Unlocking Student-side Learning in Multimodal On-Policy Distillation 提出S-OPD以解决多模态蒸馏中的学生感知瓶颈问题 distillation multimodal ✅
3 From Image Interpretation to Clinical Reasoning: Upstream Physician-Context-Aware Multimodal Learning with Causal Reinforcement Learning 提出因果强化学习框架以提升心血管事件预测能力 reinforcement learning multimodal
4 Learning to Reason with Compressed Context: Ground-Truth-Free Adaptation of OmniLLMs via Self-Distillation 提出CAFD框架以解决OmniLLMs压缩上下文推理问题 distillation privileged information large language model
5 Asking the World: Generalist Physical Reasoning through Agentic World Modeling and Probing 提出ATW以解决复杂物理推理问题 world model world models generalist agent
6 Multimodal Flow: Unified Flow Modeling of Language and Vision in Embedding Spaces 提出Multimodal Flow以解决多模态建模的量化瓶颈问题 flow matching multimodal ✅
7 LOCI: Spatial Linear Memory for Streaming World Models 提出LOCI以解决视频世界模型中的记忆与检索问题 world model world models linear attention
8 MemLife: Curating and Reasoning over Long-Term Egocentric Video Memories 提出MemLife以解决长时间自我中心视频记忆的处理问题 reinforcement learning egocentric multimodal
9 OP-CAD: On-Policy Clean-Audio Distillation for Robust Audio-Visual Reasoning 提出OP-CAD以解决环境噪声下的音频-视觉理解问题 distillation large language model multimodal
10 VR-JEPA: Learning Contrastive-State Latent Guidance for Generation-based Video Reasoning 提出VR-JEPA以解决视频生成中的推理一致性问题 JEPA Joint-Embedding Predictive Architecture joint-embedding predictive architecture
11 Physis-Lang: Self-Evolving Language as a Physical Representation for Video World Model 提出Physis-Lang以解决视频世界模型的物理一致性问题 world model world models
12 From Given to Gathered Evidence: Agentic Learning for Longitudinal Medical Reasoning 提出CASE框架以提升医学推理中的证据获取能力 reinforcement learning distillation foundation model ✅
13 MEND: Label-Free Detection, Localisation, and Correction of Latent Hallucination in World Models 提出MEND以解决世界模型中的潜在幻觉检测与修正问题 world model world models
14 Agentic Tool-Augmented Reasoning for Explainable Image Forgery Detection 提出Agentic Tool-Augmented Reasoning以解决图像伪造检测问题 curriculum learning teacher-student large language model
15 Frame Differential On-Policy Self-Distillation for Video Reasoning 提出帧差异在线自蒸馏方法以提升视频推理能力 reinforcement learning distillation multimodal
16 Emergent Multi-View Geometry Through Self-Distillation 提出Poincar3以解决多视角几何学习问题 representation learning distillation 3D reconstruction
17 Enhancing Autoregressive Video Generation via Representation Adversarial Distillation 提出Radian框架以解决自回归视频生成中的质量衰退问题 distillation foundation model
18 Diffusable Latents from Structure-Agnostic Distillation 提出无结构蒸馏方法以提升潜在扩散模型性能 distillation foundation model ✅
19 Uncertainty-Aware Consistency Distillation for Few-Step Video Generation 提出不确定性感知一致性蒸馏以解决少步视频生成问题 distillation spatiotemporal
20 ComputerSD: Online Self-Distillation from Real-Time Feedback for Computer-Use Agents 提出ComputerSD以解决计算机使用代理的在线自蒸馏问题 policy learning distillation
21 Image Classifiers are Efficient Self-Supervised Video Representation Learners 提出VideoMSN以解决视频自监督表示学习效率低的问题 representation learning foundation model ✅
22 EviRover: Reinforcing Agentic Perception Beyond a Glance 提出EviRover以解决视觉感知不足证据问题 reinforcement learning multimodal
23 Less Data, Better Timing: Student-Curriculum Coupling for VLM On-Policy Distillation in Temporal Video Grounding 提出学生-课程耦合方法以提升视频时序定位效果 distillation
24 Grounding with Confidence: Controllable Generative Video Temporal Grounding 提出可控生成视频时间定位方法以提升事件检索精度 reinforcement learning TAMP
25 From Wrecks to Wisdom: Recovering Crash Mechanics from Real-World Multi-View Photos 提出从多视角照片恢复碰撞力学以解决事故分析问题 MAE multimodal
26 Rethinking Multi-Image Re-Representation in Multi-Image Understanding 提出Mosaic以解决多图像理解中的视觉证据组织问题 reinforcement learning chain-of-thought ✅
27 Distill the Visual Evidence, Not Just the Answer: Cross-World On-Policy Distillation for Vision-Language Models 提出CW-OPD以解决视觉证据依赖不足的问题 distillation ✅
28 I Have a Stream: Making Self-Supervised Learning Work on Continuous Video 提出StreamMAE以解决连续视频自监督学习中的高相似性问题 MAE distillation
29 FrameMorrow: Future-guided Frame Selection with Prospective Tokens for Long-Horizon Video Generation 提出FrameMorrow以解决长视频生成中的历史信息选择问题 world model world models

🔬 支柱九：具身大模型 (Embodied Foundation Models) (21 篇)

#题目一句话要点标签🔗⭐
30 ShieldCLIP: Selective Safety Alignment for Harmful Content Mitigation in Multimodal Foundation Models 提出ShieldCLIP以解决多模态模型中的有害内容问题 foundation model multimodal ✅
31 GroundAnything: Reconciling Parallel Decoding with Precise Visual Grounding at Flash Speed 提出GroundAnything以解决视觉定位速度与精度的矛盾问题 foundation model visual grounding
32 Can We Anticipate Violence? Multimodal Learning from Pre-Incident Behavioral Cues 提出多模态学习方法以识别暴力事件前的行为线索 multimodal
33 Spherical Interpolation for Backward-Compatible Multimodal Representations 提出球面插值以解决多模态表示的向后兼容性问题 multimodal ✅
34 BAM! Bayesian Anything Model: a foundation model for generative computational imaging 提出BAM模型以解决生成计算成像中的物理意识缺失问题 foundation model
35 DecoMoE: Decoupling Visual Propagation and Expert Computation for Efficient Multimodal MoE Inference 提出DecoMoE以解决多模态MoE推理效率低下问题 multimodal ✅
36 NavHarness: Adaptive Goals for Agentic Vision-Language Navigation 提出NavHarness框架以解决长时间目标导航中的一致性问题 VLN multimodal
37 LongEmo: Towards Emotion Understanding and Reasoning in Long Videos 提出LongEmo以解决长视频情感理解与推理问题 large language model multimodal
38 MEMO: Multi-Level Entity-Aware Memory for Streaming Video Understanding 提出MEMO框架以解决流媒体视频理解中的记忆建模问题 large language model multimodal
39 Event-Driven Refresh and Recurrence Memory to Reduce Stale Grounding in Referring Video Object Segmentation 提出事件驱动刷新与记忆机制以解决视频物体分割中的过时问题 large language model multimodal
40 CoVisco: Codec-Native Vision Encoder with Native Token Compression for Unified Image-Video Understanding 提出CoVisco以解决长视频理解中的视觉令牌压缩问题 multimodal ✅
41 FAST: Flow Any Scene Transformer 提出FAST以解决跨视图匹配问题 foundation model
42 Rethinking Generative Image Compression at Extremely Low Bitrates 提出RAE-CoD以解决极低比特率下图像压缩问题 foundation model ✅
43 Universal Cross-Prompt Adversarial Attacks on Promptable Concept Segmentation 提出AdvPCS以解决Promptable Concept Segmentation中的对抗攻击问题 foundation model
44 TSMD: Temporal-Stream Modality Dropout for Robust Video Highlight Detection 提出TSMD以解决视频高亮检测中的模态缺失问题 multimodal
45 BadAction: Backdoor Attacks on Interactive Video Generation via Action-Guided Triggers 提出BadAction以解决交互式视频生成中的后门攻击问题 multimodal ✅
46 TED:Text-Axis Evidence Decomposition for Prompted Anomaly Localization 提出TED以解决CLIP模型在缺陷定位中的不可靠性问题 multimodal
47 When Integral Meets Decomposition: A Signal-Level Self-Supervised Feature Decompose Paradigm for Multi-Modal Image Fusion 提出信号级自监督特征分解范式以解决多模态图像融合问题 multimodal
48 Agentic Relative Camera Pose Estimation via Learned Ranking and Verification 提出PoseAgent以解决相对相机位姿估计中的多样性挑战 foundation model
49 Consensus-Aware Multi-Source Fusion for Reference-Guided Camouflaged Object Detection 提出共识感知多源融合框架以解决伪装物体检测问题 foundation model
50 Soft Spatial Reasoning 提出软空间推理以解决LVLMs的早期承诺问题 chain-of-thought ✅

🔬 支柱三：空间感知与语义 (Perception & Semantics) (9 篇)

#题目一句话要点标签🔗⭐
51 Reconstructing the Dynamic World: A Representation-Centric View of 4D Scene Reconstruction 提出统一视角以解决动态场景重建中的挑战 3D gaussian splatting 3DGS gaussian splatting ✅
52 UGOD: Uncertainty-Guided Opacity and Dropout for Sparse-View 3D Gaussian Splatting 提出UGOD框架以解决稀疏视图3D高斯渲染中的不确定性问题 3D gaussian splatting gaussian splatting splatting
53 EffGS: Efficient and High-Fidelity Gaussian Splatting 提出EffGS以解决3D高斯点云渲染效率与质量问题 3D gaussian splatting 3DGS gaussian splatting
54 EgoTools: Towards Tool-Centric Reasoning in Real-World Egocentric Videos 提出EgoTools以解决工具使用理解的挑战 affordance egocentric multimodal
55 Lens Flare Removal and Reconstruction 提出新方法以解决大光晕去除与重建问题 3DGS gaussian splatting splatting
56 GRC-Pose: Generation-Reconstruction Correspondence for Prior-Free 6D Object Pose Tracking 提出GRC-Pose以解决无先验6D物体姿态跟踪问题 sam 3D SAM 3D foundation model
57 Matisse: Evidence-Space Reasoning for Active 3D Reconstruction 提出Matisse以解决有限计算预算下的主动3D重建问题 3D reconstruction
58 Decoupling Spherical Reasoning from Dense Prediction for 360 Depth Estimation 提出Fibonacci球面图以解决全景深度估计中的几何一致性问题 depth estimation
59 Introduction to Computer Vision 提出代码优先的计算机视觉入门书籍以解决学习难题 optical flow

🔬 支柱一：机器人控制 (Robot Control) (7 篇)

#题目一句话要点标签🔗⭐
60 GroundingPI: A Grounding Foundation Model towards Physical Intelligence with Visual Primitives 提出GroundingPI以解决物理智能中的精确定位问题 manipulation reinforcement learning vision-language-action
61 Inline Memory Meets Reusable Skills: Memory-centric Framework for Vision-Language-Action Model 提出Optimus-R以解决VLA模型在新任务适应中的低效问题 manipulation vision-language-action VLA
62 Learning Skills from Historical Action Trajectories: Action Experience Dictionary for World Action Models 提出动作经验字典以解决世界动作模型的技能重用问题 manipulation world action model world action models ✅
63 StreamRig: Exploiting Intra-Rig Geometry for Streaming Multi-Camera Odometry 提出StreamRig以解决多摄像头里程计的高效利用问题 humanoid humanoid robot foundation model ✅
64 Semantic Watermarking for Malicious Image Manipulation Detection 提出语义水印技术以检测恶意图像篡改问题 manipulation
65 Uruqi: Learning Spatial Cognition from Visual Experience 提出URUQI以解决视觉语言模型空间认知不足问题 motion tracking
66 WARP: A Unified Benchmark for Invisible Image Watermarking -- Robustness and Protection Against Attacks 提出WARP基准框架以评估隐形图像水印的鲁棒性与攻击防护 manipulation ✅

🔬 支柱八：物理动画 (Physics-based Animation) (3 篇)

#题目一句话要点标签🔗⭐
67 CamAgent: An LLM-Agent Framework for Multi-Species Camera-Trap Workflows 提出CamAgent框架以解决相机陷阱数据分析碎片化问题 spatiotemporal large language model
68 TripleFlow: Training-Free Video Object Removal by Bridging Residual Editing and Native Generation 提出TripleFlow以解决视频物体移除中的背景重建问题 spatiotemporal
69 GeoGAT: Bidirectional Temporal Sampling Meets Hierarchical Graph Attention for Global Video Geo-localization 提出GeoGAT以解决全球视频地理定位中的层次冲突问题 spatiotemporal

🔬 支柱六：视频提取与匹配 (Video Extraction) (1 篇)

#题目一句话要点标签🔗⭐
70 MegaAvatar: Controllable Talking Avatar Generation 提出MegaAvatar以解决可控虚拟人生成问题 SMPL SMPL-X ✅

🔬 支柱五：交互与反应 (Interaction & Reaction) (1 篇)

#题目一句话要点标签🔗⭐
71 Perceptual Color Difference Modeling Using Machine Learning and Human Similarity Judgments 基于机器学习的人类相似性判断模型提升色差评估准确性 HSI

⬅️ 返回 cs.CV 首页 · 🏠 返回主页