cs.CV（2026-10-19）

📊 共 51 篇论文 | 🔗 8 篇有代码

🎯 兴趣领域导航

支柱二：RL算法与架构 (RL & Architecture) (18 🔗3) 支柱九：具身大模型 (Embodied Foundation Models) (14 🔗1) 支柱一：机器人控制 (Robot Control) (8 🔗1) 支柱三：空间感知与语义 (Perception & Semantics) (7 🔗2) 支柱六：视频提取与匹配 (Video Extraction) (2 🔗1) 支柱四：生成式动作 (Generative Motion) (1) 支柱五：交互与反应 (Interaction & Reaction) (1)

🔬 支柱二：RL算法与架构 (RL & Architecture) (18 篇)

#题目一句话要点标签🔗⭐
1 Think Densely, Act Sparsely: Latent Expert Cognitive Chains for Vision-Language-Action Autonomous Driving 提出LECDrive以解决自主驾驶中的稀疏监督问题 world model world models curriculum learning
2 Multi-scale Fusion of Fast Vision Mamba and Foundation Model Features for Patch-Level Gleason Grading in Computational Pathology 提出多尺度融合方法以解决前列腺癌Gleason分级问题 Mamba foundation model
3 V-JEPA Hand: Simple, Accurate and Fast 3D Hand and Camera Trajectory Recovery from Egocentric Videos 提出V-JEPA Hand以解决从自我中心视频中恢复3D手和相机轨迹的问题 JEPA egocentric hand reconstruction ✅
4 Reinforcement Learning with Visually Attributable On-Policy Self-Distillation 提出视觉可归因的在线自蒸馏强化学习以解决视觉推理中的错误定位问题 reinforcement learning distillation multimodal
5 Learning Goal-Conditioned Subgoals through Frozen Hierarchical World Models 提出轻量级目标条件网络以优化层次世界模型的规划效率 world model world models JEPA
6 See the Future, Classify the Present: Stochastic Latent World Modeling for Hyperspectral Images 提出随机潜在世界建模以解决高光谱图像分类问题 world model world models representation learning
7 PICA-WAM: Visual Perturbation Invariance via Cross-Appearance Self-Alignment in World Action Models 提出PICA-WAM以解决视觉扰动不变性问题 world action model world action models ✅
8 State World Models: Execute the Rules, Generate the Pixels 提出Stateful Harness以解决隐状态下的规则执行问题 world model world models
9 ActionRoPE: action as coordinates for spatially consistent control in planar-camera game world models 提出ActionRoPE以解决平面相机游戏世界中的动作控制问题 world model world models
10 Context-Aware Topological Distillation Graph Network for Asymmetric EEG-Eye Movement Emotion Recognition 提出上下文感知拓扑蒸馏图网络以解决不对称EEG-眼动情绪识别问题 distillation multimodal
11 FrameGRPO: Incentivizing Video Reasoning via Multi-Frame Rollouts and Temporal Credit 提出FrameGRPO以解决视频推理中的奖励同质化问题 reinforcement learning large language model multimodal
12 CheckDistill: Adaptive Checklist Distillation for Efficient Distributional Reward Modeling in Image Editing 提出CheckDistill以解决图像编辑中的奖励建模问题 reinforcement learning distillation
13 PDNet: Exploring Pruning and Distillation for Real-Time Worker Activity Recognition in Industrial Assembly 提出PDNet框架以解决工业装配中的实时人类活动识别问题 distillation
14 TriTD: Tri-Partite Trajectory-Distribution Distillation for Real-Time Autoregressive Video Generation 提出TriTD框架以解决自回归视频生成中的质量与效率问题 distillation
15 TamedFlow: Trajectory-Aware Distillation\\for Teacher-Assisted Diffusion Sampling 提出TamedFlow以解决教师辅助扩散采样中的组合差距问题 distillation
16 Functional Cross-layer Experts Routing for One-step Diffusion Distillation 提出DeMoE以解决一阶段扩散蒸馏中的内部容量结构问题 distillation ✅
17 Visual Memory and Textual Alignment for Cross-Modal Few-Shot Segmentation 提出ViTA-SAM以解决跨模态少样本分割问题 representation learning foundation model
18 Dynamics-Aware Sparse Attention for Efficient Autoregressive Video Diffusion 提出动态感知稀疏注意力以解决自回归视频扩散效率问题 distillation spatiotemporal

🔬 支柱九：具身大模型 (Embodied Foundation Models) (14 篇)

#题目一句话要点标签🔗⭐
19 MIRA: A Multimodal Illustrated Deep Research Agent for Controllable Reports 提出MIRA以解决多模态深度研究报告生成问题 foundation model multimodal
20 Scaling Vision Foundation Models for Toxicologic Pathology 提出TOX-FM以解决毒理病理学领域的模型训练问题 foundation model
21 Progressive multi-scale tumor-state memory for incomplete multimodal brain tumor segmentation 提出渐进式多尺度肿瘤状态记忆网络以解决多模态脑肿瘤分割问题 multimodal
22 LaMER: Language-Conditioned Multi-Expert Routing for Weakly Supervised Referring Expression Comprehension 提出LaMER以解决弱监督下的指称表达理解问题 language conditioned
23 HGASurv: Hierarchical Granularity-Aware Integration for Multimodal Survival Prediction 提出HGASurv以解决多模态生存预测中的粒度差异问题 multimodal
24 CheXFuse: Guided Differential Joint Attention Fusion for Robust Multimodal Chest X-Ray Classification in Data-Scarce Medical Settings 提出CheXFuse以解决多模态胸部X光分类中的信息不对称问题 multimodal
25 Beyond Omni: A Unified MLLM for Embodied Sensory Understanding 提出Unisense框架以解决多模态感知理解问题 large language model multimodal
26 Small Average Error, Wide Individual Error: General-Purpose LLMs Estimating Energy Intake from Meal Photos in a Feeding Trial 利用多模态大语言模型估算餐食能量摄入以改善饮食评估 large language model multimodal
27 DCR-TSF: Dual-Level Compositional Reasoning with Triadic Spherical Fusion for Training-Free Zero-Shot Composed Image Retrieval 提出DCR-TSF框架以解决训练无关的零-shot复合图像检索问题 large language model multimodal
28 OmniGround: Strengthening Omni-Modal Temporal Grounding via Attention-Consistency Training 提出TACG方法以解决多模态时间定位问题 large language model TAMP
29 Focused Omni-Modal GUI Agents 提出Focused Perception以解决多模态GUI代理的顺序干扰问题 multimodal
30 GeoEdit: Adversarial De-Evidencing for Silent Geoprivacy Protection via Image Editing 提出GeoEdit以解决地理隐私保护问题 multimodal
31 AVTime: Reinforcing Dense Audio-Visual Captions with Bidirectional Alignment 提出AVTime以解决多模态视频理解中的音视频密集标注问题 multimodal ✅
32 SAPO: Self-Supervised Attention Policy Optimization for Improving Temporal Video Grounding in Vision-Language Models 提出SAPO以解决视频语言模型的时间定位精度问题 TAMP

🔬 支柱一：机器人控制 (Robot Control) (8 篇)

#题目一句话要点标签🔗⭐
33 Ropedia W0: Scaling Loco-Manipulation World Models across Embodiments with Human Experience 提出Ropedia W0以解决多样化实体间的运动与操作建模问题 humanoid locomotion manipulation
34 PrismWAM: Decoupling State Representations in World Action Models 提出PrismWAM以解决世界动作模型中的状态表示耦合问题 manipulation world model world models
35 Frechet Forcing: Training Long-Horizon Streaming World Models from Short Clips 提出Fréchet Forcing以解决长时间段流媒体世界模型训练问题 manipulation flow matching world model
36 CondenseVLA: Learnable History Condensation for Efficient Multi-Frame VLA 提出CondenseVLA以解决机器人操作中的短期视觉信息整合问题 manipulation VLA ✅
37 EgoXperience: Learning World-Grounded Egocentric Representations from Human Experience 提出EgoXperience以解决人类活动重建中的一致性问题 locomotion manipulation egocentric
38 PVBench: Quantifying Physical Plausibility in Video Generation with Object-Level Physical Measurements 提出PVBench以量化视频生成中的物理合理性问题 quadruped humanoid world model
39 DenseHap: Dense Hand Haptic Estimation from Monocular Images 提出DenseHap以解决单目图像下手部触觉估计问题 manipulation hand reconstruction
40 GeoNaviDiff: Transferring Disentangled Semantic Controls to Diffusion Features via Local Riemannian Navigation 提出GeoNaviDiff以实现图像编辑的语义控制 manipulation

🔬 支柱三：空间感知与语义 (Perception & Semantics) (7 篇)

#题目一句话要点标签🔗⭐
41 Rethinking Sparse-View 3D Gaussian Splatting: A Few-Shot Learning Perspective on Overfitting Prevention 提出稀疏视图3D高斯点云重建方法以解决过拟合问题 3D gaussian splatting 3DGS gaussian splatting
42 Auditing Semantic and Relational Reliability in Open-Vocabulary Human-Object Interaction Evaluation 提出诊断协议以解决开放词汇人机交互评估中的语义和关系可靠性问题 open-vocabulary open vocabulary human-object interaction
43 Foundation Model Driven Optical Flow 提出基于基础模型的光流估计方法以提升光流精度 Depth Anything optical flow foundation model ✅
44 RxGS: Relightable Gaussian Stylization via BRDF Decomposition 提出RxGS框架以解决3D场景风格化的重光照问题 3D gaussian splatting 3DGS gaussian splatting
45 Drive2Gauss: FeedForward 4D Gaussian Reconstruction from Multimodal Generative Driving Videos 提出Drive2Gauss以解决动态4D高斯场景重建问题 metric depth spatiotemporal multimodal
46 Real-Time Radar Semantic and Panoptic Scene Understanding 提出StreamRPS以解决实时FMCW雷达场景理解问题 scene understanding
47 PnP-Transformer: A Neural Pose Solver for 6D Object Pose Estimation 提出PnP-Transformer以解决6D物体姿态估计问题 6D pose estimation ✅

🔬 支柱六：视频提取与匹配 (Video Extraction) (2 篇)

#题目一句话要点标签🔗⭐
48 ExMR: Extremity Motion Refinement for Monocular Human Motion Capture 提出ExMR以解决单目视频中手脚运动捕捉问题 hand reconstruction human motion ✅
49 Holi-World: Building Holistic World Intelligence from Human-Ego Experience 提出HOLI-WORLD以构建全面的世界智能，解决人类自我体验学习问题 egocentric

🔬 支柱四：生成式动作 (Generative Motion) (1 篇)

#题目一句话要点标签🔗⭐
50 Risk Decomposition for Hallucination-Aware Video Sampling 提出风险分解框架以解决视频生成中的幻觉问题 physically plausible

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

#题目一句话要点标签🔗⭐
51 HORIA: 4D Human and Object Reconstruction from a Monocular Video with Interaction-Aware Re-Rendering 提出HORIA以解决单目视频中人类与物体的4D重建问题 human-object interaction

⬅️ 返回 cs.CV 首页 · 🏠 返回主页