cs.CV（2026-10-20）

📊 共 61 篇论文 | 🔗 12 篇有代码

🎯 兴趣领域导航

支柱九：具身大模型 (Embodied Foundation Models) (18 🔗4) 支柱二：RL算法与架构 (RL & Architecture) (17 🔗5) 支柱三：空间感知与语义 (Perception & Semantics) (12 🔗2) 支柱一：机器人控制 (Robot Control) (9) 支柱四：生成式动作 (Generative Motion) (2 🔗1) 支柱七：动作重定向 (Motion Retargeting) (1) 支柱六：视频提取与匹配 (Video Extraction) (1) 支柱八：物理动画 (Physics-based Animation) (1)

🔬 支柱九：具身大模型 (Embodied Foundation Models) (18 篇)

#题目一句话要点标签🔗⭐
1 VGL-stGen: Learning a shared multimodal model for lung histology and spatial transcriptomics 提出VGL-stGen以解决肺组织学与空间转录组学的多模态学习问题 large language model multimodal
2 Code2Video-Compass: A Comprehensive Benchmark for General-Purpose Code-Based Video Generation across T2V, I2V, and V2V 提出Code2Video-Compass以解决代码驱动视频生成的评估不足问题 large language model multimodal instruction following
3 Brain2Vec: A Self-Supervised 3D Foundation Model for Structural Brain MRI 提出Brain2Vec以解决医学影像数据不足问题 foundation model ✅
4 RACE: Learning Multimodal Search through Referential Abstraction and Compositional Evidence 提出RACE框架以解决多模态搜索中的视觉依赖问题 multimodal
5 Towards Consistent Geospatial Representations: Aligning One Encoder to Many Foundation Models 提出GeoMixture以解决地理空间模型一致性问题 foundation model
6 GAMF-Net: Grouped Availability-Aware Multimodal Fusion for Smartphone Skin Lesion Classification 提出GAMF-Net以解决智能手机皮肤病变分类中的多模态信息融合问题 multimodal
7 PACE-MoE: Adaptive Domain Sampling and Domain-Specific LoRA for Medical Visual Grounding 提出PACE-MoE以解决医疗视觉定位中的域适应问题 visual grounding
8 Text-Anchored Modality Experts with Uncertainty-Guided Routing for Multimodal Sentiment Analysis 提出文本锚定的模态专家以解决多模态情感分析问题 multimodal
9 CountVideoBench: A Benchmark to Evaluate Prediction Biases on the Object Counting task by Video-Language Models 提出CountVideoBench以解决视频语言模型在物体计数任务中的偏差问题 large language model multimodal
10 WildAIGC-Bench: Can AI Image Detectors Handle the Wild? A Benchmark of Generated and Edited Images in Online Circulation 提出WildAIGC-Bench以解决AI生成图像检测挑战 large language model multimodal ✅
11 SynCalib: Breaking Dataset Shortcuts in LiDAR-Camera Calibration via Extrinsic-Controllable Synthesis 提出SynCalib以解决LiDAR-相机标定中的数据集捷径问题 zero-shot transfer
12 StreamEval: Evaluating Interactive Streaming Video Generation Models 提出StreamEval基准以评估交互式流媒体视频生成模型 multimodal
13 HRoPE: Homography Relative Positional Encoding 提出HRoPE以解决多视角变换中的表面几何信息缺失问题 foundation model ✅
14 PointForward V2: Point-Aligned Interaction for Feedforward Driving Reconstruction 提出PointForward V2以解决动态驾驶场景重建中的信息交互问题 TAMP
15 Matrix-Game 4.0: Harnessing Video Models for Real-Time Playable Game Worlds 提出Matrix-Game 4.0以实现实时可玩游戏世界 multimodal
16 GARRET: Generative Feedback for Region Retrieval in Visual Documents 提出GARRET框架以解决视觉文档区域检索问题 multimodal
17 A Principled Optimal Transport Framework for Frame Selection in Long Video Understanding 提出OTFS框架以解决长视频理解中的帧选择问题 multimodal ✅
18 AnomalyJev: Towards Evaluating Image Anomaly Explanations via Claim-Level Verification 提出AnomalyJev以解决图像异常解释评估问题 multimodal

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

#题目一句话要点标签🔗⭐
19 EndoWorld-JEPA: Motion-Guided Global-Local Latent World Modeling for Endoscopic Video Representation Learning 提出EndoWorld-JEPA以解决内窥镜视频表示学习中的动态建模问题 world model world models JEPA
20 PanoGraphGS: Unifying Panoptic Segmentation and Hierarchical 3D Scene Graphs with Gaussian Splatting 提出PanoGraphGS以解决3D场景图构建与全景分割问题 representation learning 3DGS gaussian splatting ✅
21 AEON: Unifying Video and 3D World Models 提出AEON框架以统一视频与3D世界模型 world model world models distillation ✅
22 Differentiable Task-Driven Coded Aperture Camera Design with 3D Gaussian Scene Representations 提出可微分任务驱动的编码光圈相机设计以优化感知性能 reinforcement learning depth estimation 3D gaussian splatting
23 OverSeer: Multi-Agent World Models Beyond What Agents See 提出OverSeer以解决多智能体世界模型一致性问题 world model world models egocentric
24 Causal Spatiotemporal State Evolution for Monocular 4D Reconstruction 提出因果时空状态演化框架以解决单目4D重建问题 Mamba representation learning optical flow
25 Convergent Performance, Divergent Mechanisms in LoRA-Adapted Vision Foundation Models for Brain MRI 提出LoRA适应性方法以优化脑MRI图像分割性能 distillation foundation model
26 Deep Learning-Based Road Lane Quality Assessment from Egocentric Dashcam Video 提出基于深度学习的车载摄像头视频车道质量评估方法 MAE scene understanding egocentric ✅
27 Touch-JEPA: Tactile Representation Learning from Hand-Object Video with Limited Tactile Supervision 提出Touch-JEPA以解决有限触觉监督下的触觉表示学习问题 JEPA representation learning
28 World Models with Configurable Perception 提出ConfigWM以解决世界建模中的信息干扰问题 world model world models
29 Answering in the User's Variety: Calibrated Accommodation Pairs for Post-Training 提出校准适应对以解决用户方言匹配问题 reinforcement learning DPO instruction following ✅
30 Decoupled Representation Learning for Face Presentation Attack Detection under Smart Glass Capture 提出SmaGla-FPAD以解决智能眼镜捕获下的人脸伪造攻击检测问题 representation learning
31 TokenFlow: Latent Token Rectified Flow Matching for 3D Point Cloud Generation 提出TokenFlow以解决3D点云生成中的结构保持问题 flow matching
32 CLOSER: Evidence-Aware Active Visual Verification for Embodied Object-Goal Navigation 提出CLOSER以解决物体目标导航中的在线验证问题 diffusion policy egocentric
33 Latent Bridge: From Video Latents to 3D Geometry via Feature Alignment 提出Latent Bridge以解决视频生成与3D重建的连接问题 distillation 3D reconstruction ✅
34 SplatFlux: Feed-Forward 4D Gaussian Reconstruction and Generation from Video Latents 提出SplatFlux以解决视频基础4D重建与生成问题 distillation optical flow
35 WildMAE: Structure-Aware Masked Autoencoding for Wildlife Image Reconstruction 提出WildMAE以解决野生动物图像重建问题 masked autoencoder MAE

🔬 支柱三：空间感知与语义 (Perception & Semantics) (12 篇)

#题目一句话要点标签🔗⭐
36 CoGaussian: Open-Vocabulary Gaussian-based Collaborative 3D Scene Understanding 提出CoGaussian以解决协作感知中的语义占用预测问题 scene understanding open-vocabulary open vocabulary
37 Learning from Automatic 3D Gaussian Splatting Edits for Online Bayesian Scene Change Detection 提出基于自动3D高斯点云编辑的在线贝叶斯场景变化检测方法 3D gaussian splatting 3DGS gaussian splatting
38 CrossFit-GS: Out-of-Fold Pose Estimation for COLMAP-Free 3D Gaussian Splatting 提出CrossFit-GS以解决3D姿态估计中的轨迹漂移问题 3D gaussian splatting 3DGS gaussian splatting
39 RLSplat: Resource-Linear Budget Control for Feed-forward 3D Gaussian Splatting 提出RLSplat以解决FF3DGS推理资源控制问题 3D gaussian splatting gaussian splatting splatting
40 KoalaGS: Complexity-Aware Continuous Level of Detail for Gaussian Splatting 提出KoalaGS以解决高斯点云渲染的复杂性问题 3D gaussian splatting 3DGS gaussian splatting
41 FocusOVD: Geometric and Contextual Feature Refinement for Open-Vocabulary Small-Object Detection 提出FocusOVD以解决小物体开放词汇检测问题 open-vocabulary open vocabulary
42 HiQGS: Hierarchical Spatiotemporal Querying for Feed-Forward Gaussian Reconstruction 提出HiQGS以解决稀疏单目帧的高斯重建问题 scene reconstruction spatiotemporal TAMP
43 OV-Foam: How 3D Representations Shape Semantic Feature Lifting 提出OV-Foam以解决3D特征提升中的重叠问题 3D gaussian splatting gaussian splatting splatting
44 Geometry Regularization for Vanilla NeRF via Self-Supervised Cross-Ray Consistency 提出几何正则化方法以解决NeRF几何不准确问题 NeRF neural radiance field
45 OmniVLoc: Topo-metric Visual Localization for Omnidirectional Cameras 提出OmniVLoc以解决全景相机视觉定位问题 depth estimation monocular depth
46 Neural Implicit Energy Optimization for LiDAR Mapping 提出神经隐式能量优化以改善LiDAR映射的平滑性 implicit representation ✅
47 VEOcc++: Learning Persistent Voxel Feature Maps for Embodied Online Semantic Occupancy Mapping and Offline Refinement 提出VEOcc++以解决在线增量语义占用映射与离线优化问题 scene understanding ✅

🔬 支柱一：机器人控制 (Robot Control) (9 篇)

#题目一句话要点标签🔗⭐
48 World2HOSI: A Multi-Expert Harness for Long-Horizon Human-Object-Scene Interaction Generation 提出World2HOSI以解决长时间人-物-场景交互生成问题 manipulation 3D gaussian splatting 3DGS
49 RoboJudge: Multimodal Language Models as Judges for Embodied Video Generation 提出RoboJudge以评估视频生成模型的物理有效性与任务一致性 dexterous hand reinforcement learning world model
50 VLA-Retain: Benchmarking Cross-Dimension Effects of Vision Tuning in VLAs 提出VLA-Retain以评估视觉调优在VLA中的跨维度影响 manipulation vision-language-action VLA
51 LaLa-World: Scaling Latent World Models with Natural Language Actions 提出LaLa-World以提升多领域规划与策略优化 MPC world model world models
52 Decoupling Action from Egocentric Observation for World Simulation 提出EgoACT以解决现有自我中心动作转移中的几何估计误差问题 manipulation policy learning physically plausible
53 Media Intelligence Platform: Multimodal Detection of Synthetic and Manipulated Content 提出媒体智能平台以解决多模态内容检测问题 manipulation multimodal
54 Learning the 4D World through Point Trajectory Forecasting 提出视频点模型以解决多视角视频预测问题 manipulation world model world models
55 NeoWorld-3: From Pixels to Code-Defined Physical Worlds 提出NeoWorld-3以解决3D世界重建中的物理交互问题 manipulation dual-arm
56 Reusing Assurance Evidence in AI Pipelines: Validity Conditions, a Fail-Closed Release Gate, and a Replay of 714 Public Data Version Control Pipelines 提出重用AI管道中的保证证据以提高效率 manipulation

🔬 支柱四：生成式动作 (Generative Motion) (2 篇)

#题目一句话要点标签🔗⭐
57 Learn to Modify: Self-evolving Multi-agent Refinement Framework for Conversational Text-to-Motion Generation 提出自演化多智能体框架以解决文本到动作生成中的迭代反馈问题 text-to-motion motion generation spatiotemporal
58 Gaze2HOI: Gaze-Aware Intention Disambiguation for Egocentric Hand-Object Interaction Forecasting 提出Gaze2HOI以解决手-物体交互意图模糊问题 motion generation egocentric ✅

🔬 支柱七：动作重定向 (Motion Retargeting) (1 篇)

#题目一句话要点标签🔗⭐
59 PI-MLLM: Psychophysics-Inspired Multimodal Large Language Model for Oddly Satisfying Video Emotion Analysis 提出PI-MLLM以解决Oddly Satisfying视频情感分析问题 motion prediction large language model multimodal

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

#题目一句话要点标签🔗⭐
60 GazeLens for Efficient Egocentric Video Question Answering with Simple Gaze-Driven Adaptive Token Reduction 提出GazeLens以解决有限计算下的自我中心视频问答问题 egocentric

🔬 支柱八：物理动画 (Physics-based Animation) (1 篇)

#题目一句话要点标签🔗⭐
61 Recast the Past: Stable Long Video Generation with Full Context Memory 提出ReCam以解决长视频生成中的历史信息保留问题 spatiotemporal

⬅️ 返回 cs.CV 首页 · 🏠 返回主页