cs.CV（2026-09-16）

📊 共 27 篇论文 | 🔗 3 篇有代码

🎯 兴趣领域导航

支柱三：空间感知与语义 (Perception & Semantics) (6) 支柱九：具身大模型 (Embodied Foundation Models) (6) 支柱二：RL算法与架构 (RL & Architecture) (5 🔗1) 支柱八：物理动画 (Physics-based Animation) (5 🔗1) 支柱一：机器人控制 (Robot Control) (4 🔗1) 支柱六：视频提取与匹配 (Video Extraction) (1)

🔬 支柱三：空间感知与语义 (Perception & Semantics) (6 篇)

#题目一句话要点标签🔗⭐
1 NormLift: From Lifted Features To Semantic Reliability In 3D Gaussian Splatting 提出NormLift以解决3D高斯特征语义可靠性问题 3D gaussian splatting gaussian splatting splatting
2 Geometry beneath the Waves: Dense Priors for Sparse-View Underwater 3D Gaussian Splatting 提出稀疏视图下的密集先验以解决水下3D重建问题 3D gaussian splatting 3D reconstruction gaussian splatting
3 CADSplat: Sparse-View 3D Gaussian Splatting Aided by CAD Models for Robust, Photorealistic Digital-Twin Reconstruction 提出CADSplat框架以解决稀疏视图下数字双胞胎重建问题 3D gaussian splatting 3DGS gaussian splatting
4 GeoCond: A Conditioning-Aware Reliability Adapter for Feed-Forward 3D Reconstruction 提出GeoCond以解决低重叠和极端旋转下的3D重建可靠性问题 3D reconstruction VGGT foundation model
5 CapMap-MS-TTA: 3rd Place Solution for the MUMU Track of the 8th LSVOS Challenge at ECCV 2026 提出CapMap-MS-TTA以解决多模态视频对象分割问题 open-vocabulary open vocabulary multimodal
6 Finder: Agentic Closed-Loop Object Finding for Embodied Grounding 提出Finder以解决3D场景中物体查找问题 open-vocabulary open vocabulary VLN

🔬 支柱九：具身大模型 (Embodied Foundation Models) (6 篇)

#题目一句话要点标签🔗⭐
7 FIVE-VLA: Fast and EffectIVE Autonomous Driving with Recurrent Action Memory 提出FIVE-VLA以解决自主驾驶中的高参数与记忆不足问题 vision-language-action VLA
8 Generalist-Specialist Mixture-of-Experts for Rare Pathology Detection in Multimodal Imaging 提出通用-专用混合专家模型以解决稀有病理检测问题 multimodal
9 MCLC-NET: Multimodal Continual Learning for Leaf Counting 提出MCLC-NET以解决多模态持续学习的叶片计数问题 multimodal
10 In-Context Robot Learning with VLM Agents 提出GPT-Policy以解决机器人在新环境中的适应性问题 embodied AI
11 RankGround: Efficient High-Resolution GUI Grounding via Lightweight Reranker-Guided Crop Selection 提出RankGround以解决高效高分辨率GUI定位问题 multimodal
12 Semantic-ITC: A Frame-wise Indoor Mobile Laser Scanning Dataset and Benchmark for Semantic Segmentation 提出Semantic-ITC数据集以解决室内激光扫描语义分割问题 foundation model

🔬 支柱二：RL算法与架构 (RL & Architecture) (5 篇)

#题目一句话要点标签🔗⭐
13 CSWAM: Better Causal Semantic Representations for Out-of-Distribution Generalization in World Action Models 提出CSWAM以解决视觉分布转移下的因果语义表示问题 world action model world action models JEPA
14 StrucPhysVideo: Learning Physical Dynamics from Structured Captions and Robot Actions 提出StrucPhysVideo以解决物理动态建模问题 world model world models distillation
15 Mask 2D-3D: Adaptive Dual-Masked Autoencoder Network for Image-to-Point Cloud Registration 提出自适应双重掩码自编码器网络以解决图像与点云配准问题 reinforcement learning representation learning masked autoencoder
16 Can MiniMax-H3 Reason About the Physical World? An Evaluation of Omni-Modal Generative Model 提出MiniMax-H3以解决多模态世界推理问题 world model world models multimodal ✅
17 Vocabulary-Guided Gait Recognition 提出词汇引导的步态识别方法以提升步态理解 world model world models

🔬 支柱八：物理动画 (Physics-based Animation) (5 篇)

#题目一句话要点标签🔗⭐
18 PULSE: Unlocking Practical Image Compression on Single-Thread CPU 提出PULSE以解决单线程CPU上的图像压缩效率问题 PULSE ✅
19 STUNet-Fusion: Spatiotemporal Needle-Tip Localization in Ultrasound Video via Multi-Channel Motion Fusion 提出STUNet-Fusion以解决超声视频中针尖定位问题 spatiotemporal
20 Mask IPL: Noise-Free Intrinsic Position Learning via Computation Graph Clipping for Event-Based Spike-Driven Tracking 提出Mask IPL以解决事件驱动跟踪中的噪声问题 spatiotemporal
21 VibeAvatar: Aligning Phonetic Kinematics and Human Aesthetics for High-Fidelity Talking Avatar Synthesis 提出VibeAvatar以解决多模态虚拟人合成中的准确性与美学问题 AMP
22 vidax: A Unified JAX Framework for Video Generative Models on Accelerator Meshes 提出vidax框架以解决TPU视频生成模型推理问题 spatiotemporal

🔬 支柱一：机器人控制 (Robot Control) (4 篇)

#题目一句话要点标签🔗⭐
23 HAP: A Hand-Driven Active Perception Framework for Egocentric Head Motion Prediction 提出HAP框架以解决人类头部运动预测问题 manipulation egocentric motion prediction ✅
24 Sim-to-Real Traffic Scene Understanding by Decoupling Semantics from Caption Generation with V-JEPA 提出解耦语义与描述生成的框架以解决交通场景理解问题 sim-to-real JEPA scene understanding
25 PointZero: 3D Point Track Completion for Learning Transferable 3D Dynamics 提出PointZero以解决无机器人数据的3D动态学习问题 manipulation imitation learning world model
26 Track, Articulate, Act: Generating Articulation from Casual Human Videos 提出一种新方法从人类视频生成可动对象模型以解决机器人交互问题 manipulation 3D reconstruction scene flow

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

#题目一句话要点标签🔗⭐
27 Video-Based Markerless Motion Capture for Clinical and Rehabilitation Biomechanics: A PRISMA-ScR Scoping Review of Validated Architectures, Clinical Readiness, and Emerging Methods 评估视频无标记运动捕捉在临床生物力学中的应用潜力 markerless motion capture foundation model

⬅️ 返回 cs.CV 首页 · 🏠 返回主页