cs.CV（2026-09-25）

📊 共 35 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) (13) 支柱二：RL算法与架构 (RL & Architecture) (12 🔗1) 支柱一：机器人控制 (Robot Control) (4 🔗1) 支柱九：具身大模型 (Embodied Foundation Models) (4) 支柱四：生成式动作 (Generative Motion) (1) 支柱七：动作重定向 (Motion Retargeting) (1 🔗1)

🔬 支柱三：空间感知与语义 (Perception & Semantics) (13 篇)

#题目一句话要点标签🔗⭐
1 DepthEvidence: Unifying Metric Depth Prediction and Geometric Reasoning in Multimodal Language Models 提出DepthEvidence以统一度量深度预测与几何推理 depth estimation metric depth multimodal
2 ClearGS: Reliability-Aware Gaussian Splatting from Handheld Videos 提出ClearGS以解决手持视频中的3D高斯点云重建问题 3D gaussian splatting 3DGS gaussian splatting
3 Gauss What You Need: Compact Gaussian Splatting Across Scene Scales 提出TangoGS以解决3D高斯点云重建的规模适应问题 3D gaussian splatting gaussian splatting splatting
4 OC-GS: Gaussian Splatting for Irregular Turntable Capture 提出OC-GS以解决不均匀转台捕获重建问题 gaussian splatting splatting
5 Open Vocabulary Domain Unlearning 提出开放词汇领域遗忘方法以解决视觉语言模型的偏见问题 open-vocabulary open vocabulary
6 CytoSPM: Open-Vocabulary Cytopathology Detection with Structured Prompt Bank 提出CytoSPM以解决细胞病理学开放词汇检测问题 open-vocabulary open vocabulary
7 Self-Supervised Perceptually Interpretable Monocular Depth Estimation 提出自监督感知可解释的单目深度估计方法以提升透明度 depth estimation monocular depth
8 From Mono to Stereo: Accelerating Binocular Gaussian Splatting via Reprojection and Selective Patching 提出双目高斯点云加速方法以解决渲染效率问题 gaussian splatting splatting
9 Enhancing Photogrammetric Digital Surface Models with Pretrained Diffusion Models and Multimodal Conditioning 提出多模态条件下的扩散模型以提升数字表面模型精度 elevation map multimodal
10 Spackle: Completing Large View Single Image NVS with Adaptive Gaussians 提出Spackle以解决大视角单图像新视角合成中的容量竞争问题 3D gaussian splatting 3DGS gaussian splatting
11 GraphWrit3R: End-to-End 3D Scene Graph Writing 提出GraphWrit3R以解决3D场景图生成中的多重限制问题 open-vocabulary open vocabulary large language model
12 DAPEVO: Deep Adaptive Patch Frame-Event Visual Odometry 提出DAPEVO以解决视觉里程计在动态环境中的鲁棒性问题 visual odometry
13 Double-stream registration with pyramid fusion for HDR video with alternating exposures 提出双流注册与金字塔融合以解决HDR视频重建问题 optical flow

🔬 支柱二：RL算法与架构 (RL & Architecture) (12 篇)

#题目一句话要点标签🔗⭐
14 Timo: $\textbf{T}$aming Mult$\textbf{i}$modal Diffusion Transformer for Human $\textbf{Mo}$tion Generation 提出Timo框架以解决人类动作生成中的多模态建模问题 flow matching motion generation human motion
15 DyMD: Preserving Interaction Dynamics through Distribution Matching Distillation in Few-Step Video World Models 提出DyMD以解决视频生成中的动态交互问题 world model world models distillation
16 MM-VeriAgent: Learning to Use Extensive Tools to Verify Multimodal Misinformation with Reinforcement Learning 提出MM-VeriAgent以解决多模态虚假信息验证问题 reinforcement learning multimodal
17 KneePreM: Towards 3D Knee MRI Foundation Models via Large-Scale Unlabeled Pretraining and Label-Efficient Fine-Tuning 提出KneePreM以提高膝关节MRI分类与分割的标签效率 masked autoencoder foundation model
18 Structured Reasoning Agentic Framework for Interpretable Critical View of Safety Assessment 提出ReasonCVS框架以解决腹腔镜胆囊切除术中的安全视图评估问题 distillation scene understanding spatial relationship
19 ContraFM-S2O: Flow Matching-Based One-step SAR-to-Optical Image Translation Model with Contrastive Learning 提出ContraFM-S2O以解决SAR到光学图像转换中的细节缺失问题 flow matching contrastive learning
20 OneWorld: Learning Consistent Physics Across Actions in World Models 提出OneWorld以解决动态环境中物理一致性问题 world model world models
21 StarWM: Self-Supervised Trained Attention Routing for Robust World Models 提出StarWM以解决世界模型中信息重构与动态相关性平衡问题 world model world models
22 FlatClip: A Geometry-Aware Surface-Level Baseline for fMRI Representation Learning 提出FlatClip以解决fMRI表示学习中的空间尺度问题 representation learning foundation model ✅
23 Structure-Guided Masked Autoencoders for Ultra-High Resolution Scientific Image Understanding 提出结构引导的掩码自编码器以解决超高分辨率科学图像理解问题 masked autoencoder MAE
24 MVVBench: Benchmarking 4D Reasoning in Vision-Language Models 提出MVVBench以解决多视角视频理解中的4D推理问题 reinforcement learning chain-of-thought
25 Conditional Predictive Sufficient Statistics for Visual Representation Learning 提出条件预测充分统计量以提升视觉表征学习效果 representation learning

🔬 支柱一：机器人控制 (Robot Control) (4 篇)

#题目一句话要点标签🔗⭐
26 Reliability-Regulated Trajectory Optimization for Progressive COLMAP-Free 3D Gaussian Splatting 提出可靠性调节的轨迹优化以解决COLMAP-free 3D Gaussian Splatting问题 trajectory optimization 3D gaussian splatting 3DGS ✅
27 TRACKGRAPH: Online Open-Vocabulary 3D Scene Graphs via Image-Space Tracking 提出TRACKGRAPH以解决在线开放词汇3D场景图构建问题 quadruped open-vocabulary open vocabulary
28 ManiVid: Unified and Explainable Forensic Analysis of Manipulated Videos 提出ManiVid以解决视频伪造分析的挑战 manipulation open-vocabulary open vocabulary
29 AxonSynth: Domain-Randomized Synthetic Data for Zero-Shot 3D Axon Segmentation in Light-Sheet Microscopy 提出AxonSynth以解决3D显微镜中轴突分割的标注依赖问题 domain randomization zero-shot transfer

🔬 支柱九：具身大模型 (Embodied Foundation Models) (4 篇)

#题目一句话要点标签🔗⭐
30 Skip the Talk, Re-Focus on Vision: Latent Reasoning for Reasoning Segmentation in Multimodal Large Language Models 提出LIRSeg以解决多模态大语言模型中的推理分割问题 large language model multimodal chain-of-thought
31 VLALight: Lightweight Vision-Language-Action Models for Emergency-Aware Traffic Signal Control 提出VLALight以解决紧急车辆交通信号控制问题 vision-language-action
32 Pocket-STVG: lightweight architecture for Spatio-Temporal Video Grounding 提出Pocket-STVG以解决轻量级时空视频定位问题 large language model multimodal
33 OpenVAM: Open-World Visual Attention Modeling with VLMs 提出OpenVAM以解决视觉注意力建模中的解释性与通用性问题 instruction following

🔬 支柱四：生成式动作 (Generative Motion) (1 篇)

#题目一句话要点标签🔗⭐
34 Motion Style Slider: Endpoint-Supervised Continuous Style Control for Human Motion Diffusion 提出运动风格滑块以解决人类动作风格控制问题 motion diffusion motion generation human motion

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

#题目一句话要点标签🔗⭐
35 Geometric Inconsistency Localization in Multi-View Image Sets 提出DEFECt3R以解决多视图图像集中的几何不一致性定位问题 geometric consistency ✅

⬅️ 返回 cs.CV 首页 · 🏠 返回主页