cs.CV（2025-11-12）

📊 共 22 篇论文 | 🔗 7 篇有代码

🎯 兴趣领域导航

支柱三：空间感知 (Perception & SLAM) (14 🔗3) 支柱二：RL算法与架构 (RL & Architecture) (4 🔗2) 支柱一：机器人控制 (Robot Control) (2 🔗1) 支柱五：交互与反应 (Interaction & Reaction) (1 🔗1) 支柱六：视频提取与匹配 (Video Extraction & Matching) (1)

🔬 支柱三：空间感知 (Perception & SLAM) (14 篇)

#	题目	一句话要点	标签	🔗	⭐
1	DualVision ArthroNav: Investigating Opportunities to Enhance Localization and Reconstruction in Image-based Arthroscopy Navigation via External Cameras	DualVision ArthroNav：利用外部相机增强图像引导关节镜导航的定位与重建	visual odometry SLAM scene reconstruction
2	PALMS+: Modular Image-Based Floor Plan Localization Leveraging Depth Foundation Model	提出PALMS+以解决室内定位精度不足问题	depth estimation monocular depth point cloud	✅
3	OUGS: Active View Selection via Object-aware Uncertainty Estimation in 3DGS	OUGS：基于对象感知不确定性估计的3DGS主动视角选择	3D gaussian splatting 3DGS gaussian splatting
4	STORM: Segment, Track, and Object Re-Localization from a Single Image	提出STORM，无需人工标注，实现单图像的物体分割、跟踪和重定位。	pose estimation localization feature matching
5	DreamPose3D: Hallucinative Diffusion with Prompt Learning for 3D Human Pose Estimation	DreamPose3D：结合提示学习的幻觉扩散模型用于3D人体姿态估计	pose estimation 3D pose estimation
6	BronchOpt : Vision-Based Pose Optimization with Fine-Tuned Foundation Models for Accurate Bronchoscopy Navigation	BronchOpt：基于视觉和微调基础模型的支气管镜导航位姿优化	localization navigation
7	Spatio-Temporal Data Enhanced Vision-Language Model for Traffic Scene Understanding	提出ST-CLIP模型，利用时空信息增强视觉-语言模型，用于交通场景理解。	scene understanding navigation
8	EPSegFZ: Efficient Point Cloud Semantic Segmentation for Few- and Zero-Shot Scenarios with Language Guidance	提出EPSegFZ，利用语言引导实现高效的点云少样本/零样本语义分割	point cloud
9	OG-PCL: Efficient Sparse Point Cloud Processing for Human Activity Recognition	提出OG-PCL网络，用于高效处理稀疏雷达点云的人体活动识别	point cloud
10	Task-Aware 3D Affordance Segmentation via 2D Guidance and Geometric Refinement	提出TASA框架，融合2D引导与几何优化，实现任务感知的3D可交互区域分割	affordance detection point cloud
11	RadHARSimulator V2: Video to Doppler Generator	RadHARSimulator V2：提出一种视频到多普勒谱的雷达人体活动识别模拟器。	pose estimation PULSE	✅
12	HOTFLoc++: End-to-End Hierarchical LiDAR Place Recognition, Re-Ranking, and 6-DoF Metric Localisation in Forests	HOTFLoc++：森林环境下端到端分层LiDAR定位与重排序	point cloud
13	PIFF: A Physics-Informed Generative Flow Model for Real-Time Flood Depth Mapping	提出PIFF模型以解决实时洪水深度映射问题	depth estimation
14	Neural B-frame Video Compression with Bi-directional Reference Harmonization	提出BRHVC，通过双向参考帧协调优化神经B帧视频压缩性能	optical flow	✅

🔬 支柱二：RL算法与架构 (RL & Architecture) (4 篇)

#	题目	一句话要点	标签	🔗	⭐
15	SasMamba: A Lightweight Structure-Aware Stride State Space Model for 3D Human Pose Estimation	SasMamba：轻量级结构感知步幅状态空间模型，用于3D人体姿态估计	Mamba SSM state space model	✅
16	PAN: A World Model for General, Interactable, and Long-Horizon World Simulation	PAN：通用、可交互、长时程世界模拟的世界模型	world model latent dynamics
17	Learning by Neighbor-Aware Semantics, Deciding by Open-form Flows: Towards Robust Zero-Shot Skeleton Action Recognition	提出Flora，通过邻域感知语义和开放式流解决鲁棒的零样本骨骼动作识别问题	flow matching geometric consistency	✅
18	4KDehazeFlow: Ultra-High-Definition Image Dehazing via Flow Matching	提出4KDehazeFlow，通过Flow Matching实现超高清图像去雾，提升色彩保真度。	flow matching

🔬 支柱一：机器人控制 (Robot Control) (2 篇)

#	题目	一句话要点	标签	🔗	⭐
19	Understanding the Representation of Older Adults in Motion Capture Locomotion Datasets	分析MoCap老年人运动数据集，揭示现有数据集对老年人步态表征的不足	locomotion gait walking
20	RF-DETR: Neural Architecture Search for Real-Time Detection Transformers	RF-DETR：面向实时目标检测Transformer的神经架构搜索	running	✅

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

#	题目	一句话要点	标签	🔗	⭐
21	PressTrack-HMR: Pressure-Based Top-Down Multi-Person Global Human Mesh Recovery	PressTrack-HMR：提出基于压力感知的多人全局人体网格重建方法	multi-person interaction human mesh recovery HMR	✅

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

#	题目	一句话要点	标签	🔗	⭐
22	Enriching Knowledge Distillation with Cross-Modal Teacher Fusion	提出RichKD，通过跨模态CLIP知识融合提升知识蒸馏效果	feature matching

⬅️ 返回 cs.CV 首页 · 🏠 返回主页