cs.CV（2025-06-30）

📊 共 43 篇论文 | 🔗 16 篇有代码

🎯 兴趣领域导航

支柱二：RL算法与架构 (RL & Architecture) (13 🔗4) 支柱三：空间感知与语义 (Perception & Semantics) (12 🔗4) 支柱九：具身大模型 (Embodied Foundation Models) (11 🔗5) 支柱一：机器人控制 (Robot Control) (3 🔗1) 支柱四：生成式动作 (Generative Motion) (1) 支柱七：动作重定向 (Motion Retargeting) (1) 支柱八：物理动画 (Physics-based Animation) (1 🔗1) 支柱六：视频提取与匹配 (Video Extraction) (1 🔗1)

🔬 支柱二：RL算法与架构 (RL & Architecture) (13 篇)

#题目一句话要点标签🔗⭐
1 VOCAL: Visual Odometry via ContrAstive Learning 提出VOCAL框架以解决视觉里程计的可解释性问题 representation learning contrastive learning visual odometry
2 Mamba-FETrack V2: Revisiting State Space Model for Frame-Event based Visual Object Tracking 提出Mamba-FETrack V2以解决多模态视觉目标跟踪效率问题 Mamba state space model multimodal ✅
3 JAM-Flow: Joint Audio-Motion Synthesis with Flow Matching 提出JAM-Flow以解决音频与面部动作合成问题 flow matching motion synthesis
4 Embedding-based Retrieval in Multimodal Content Moderation 提出嵌入式检索方法以解决短视频内容审核效率问题 contrastive learning multimodal
5 Towards foundational LiDAR world models with efficient latent flow matching 提出基于潜在条件流匹配的LiDAR世界模型以解决领域迁移问题 flow matching world model
6 Dataset Distillation via Vision-Language Category Prototype 提出视觉-语言类别原型的蒸馏方法以提升数据集蒸馏性能 distillation large language model ✅
7 LLM-enhanced Action-aware Multi-modal Prompt Tuning for Image-Text Matching 提出LLM增强的动作感知多模态提示调优以解决图像-文本匹配问题 representation learning spatial relationship large language model
8 NavMorph: A Self-Evolving World Model for Vision-and-Language Navigation in Continuous Environments 提出NavMorph以解决视觉语言导航中的环境适应问题 world model VLN ✅
9 CS-VLM: Compressed Sensing Attention for Efficient Vision-Language Representation Learning 提出压缩感知注意力机制以解决视觉语言模型的计算瓶颈问题 representation learning multimodal
10 Room Scene Discovery and Grouping in Unstructured Vacation Rental Image Collections 提出房间场景发现与分组方法以解决度假租赁图像无结构问题 contrastive learning large language model
11 FADRM: Fast and Accurate Data Residual Matching for Dataset Distillation 提出FADRM以解决数据蒸馏中的信息消失问题 distillation ✅
12 From Sight to Insight: Unleashing Eye-Tracking in Weakly Supervised Video Salient Object Detection 提出基于眼动追踪的弱监督视频显著目标检测方法 contrastive learning spatiotemporal
13 When Test-Time Adaptation Meets Self-Supervised Models 提出自监督测试时适应协议以提升模型性能 contrastive learning distillation

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

#题目一句话要点标签🔗⭐
14 AttentionGS: Towards Initialization-Free 3D Gaussian Splatting via Structural Attention 提出AttentionGS以解决3D重建中对高质量点云的依赖问题 3D gaussian splatting 3DGS gaussian splatting
15 PGOV3D: Open-Vocabulary 3D Semantic Segmentation with Partial-to-Global Curriculum 提出PGOV3D以解决开放词汇3D语义分割中的信息转移问题 open-vocabulary open vocabulary large language model
16 MILo: Mesh-In-the-Loop Gaussian Splatting for Detailed and Efficient Surface Reconstruction 提出MILo框架以解决高质量3D表面重建问题 gaussian splatting splatting
17 Can We Challenge Open-Vocabulary Object Detectors with Generated Content in Street Scenes? 利用生成内容挑战开放词汇物体检测器的局限性 open-vocabulary open vocabulary
18 Instant GaussianImage: A Generalizable and Self-Adaptive Image Representation via 2D Gaussian Splatting 提出自适应高斯图像表示框架以解决训练效率低下问题 gaussian splatting splatting
19 Diffusion-Based Image Augmentation for Semantic Segmentation in Outdoor Robotics 提出基于扩散的图像增强方法以解决户外机器人语义分割问题 open-vocabulary open vocabulary foundation model
20 TextMesh4D: Text-to-4D Mesh Generation via Jacobian Deformation Field 提出TextMesh4D以解决动态3D网格生成问题 3DGS NeRF spatiotemporal
21 PriOr-Flow: Enhancing Primitive Panoramic Optical Flow with Orthogonal View 提出PriOr-Flow以解决全景光流估计中的极区失真问题 optical flow ✅
22 Proteus-ID: ID-Consistent and Motion-Coherent Video Customization 提出Proteus-ID以解决视频身份一致性与运动连贯性问题 optical flow multimodal ✅
23 Computer Vision for Objects used in Group Work: Challenges and Opportunities 提出FiboSB数据集以解决协作任务中的6D姿态估计问题 6D pose estimation
24 C3VDv2 -- Colonoscopy 3D video dataset with enhanced realism 提出C3VDv2数据集以解决3D结肠镜重建算法训练不足问题 optical flow ✅
25 SCORP: Scene-Consistent Object Refinement via Proxy Generation and Tuning 提出SCORP以解决场景重建中对象视角缺失问题 scene reconstruction ✅

🔬 支柱九：具身大模型 (Embodied Foundation Models) (11 篇)

#题目一句话要点标签🔗⭐
26 A Survey on Vision-Language-Action Models for Autonomous Driving 综述视觉-语言-动作模型以推动自动驾驶技术发展 vision-language-action VLA large language model ✅
27 Evaluation of Geolocation Capabilities of Multimodal Large Language Models and Analysis of Associated Privacy Risks 评估多模态大语言模型的地理定位能力以应对隐私风险 large language model multimodal ✅
28 Unified Multimodal Understanding via Byte-Pair Visual Encoding 提出统一多模态理解框架以解决模态对齐问题 large language model foundation model multimodal
29 Foundation Models for Zero-Shot Segmentation of Scientific Images without AI-Ready Data 提出Zenesis以解决科学图像零-shot分割问题 foundation model multimodal
30 DenseWorld-1M: Towards Detailed Dense Grounded Caption in the Real World 提出DenseWorld-1M以解决现有图像描述数据集缺乏细节的问题 large language model multimodal visual grounding
31 Towards an Automated Multimodal Approach for Video Summarization: Building a Bridge Between Text, Audio and Facial Cue-Based Summarization 提出一种多模态视频摘要方法以提升视频内容理解 multimodal
32 Flash-VStream: Efficient Real-Time Understanding for Long Video Streams 提出Flash-VStream以解决长视频理解的效率问题 large language model multimodal ✅
33 On the Domain Robustness of Contrastive Vision-Language Models 提出Deepbench框架以评估视觉-语言模型的领域鲁棒性 large language model foundation model
34 VAP-Diffusion: Enriching Descriptions with MLLMs for Enhanced Medical Image Generation 提出VAP-Diffusion以解决医学图像生成中的描述不足问题 large language model chain-of-thought
35 Learning Frequency and Memory-Aware Prompts for Multi-Modal Object Tracking 提出频率与记忆感知提示以解决多模态目标跟踪问题 foundation model ✅
36 AI-Generated Lecture Slides for Improving Slide Element Detection and Retrieval 提出SynLecSlideGen以解决讲义幻灯片元素检测与检索问题 large language model ✅

🔬 支柱一：机器人控制 (Robot Control) (3 篇)

#题目一句话要点标签🔗⭐
37 Thinking with Images for Multimodal Reasoning: Foundations, Methods, and Future Frontiers 提出以图像思维推动多模态推理的框架以解决语义差距问题 manipulation multimodal chain-of-thought
38 Epona: Autoregressive Diffusion World Model for Autonomous Driving 提出Epona以解决自主驾驶中的长时序预测问题 motion planning world model spatiotemporal ✅
39 SG-LDM: Semantic-Guided LiDAR Generation via Latent-Aligned Diffusion 提出SG-LDM以解决激光雷达点云生成问题 manipulation

🔬 支柱四：生成式动作 (Generative Motion) (1 篇)

#题目一句话要点标签🔗⭐
40 MotionGPT3: Human Motion as a Second Modality 提出MotionGPT3以解决多模态运动理解与生成问题 motion generation MotionGPT large language model

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

#题目一句话要点标签🔗⭐
41 A Unified Framework for Stealthy Adversarial Generation via Latent Optimization and Transferability Enhancement 提出统一框架以解决扩散模型对抗样本生成的转移性问题 latent optimization

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

#题目一句话要点标签🔗⭐
42 Ella: Embodied Social Agents with Lifelong Memory 提出Ella以解决社交智能体的终身学习问题 spatiotemporal foundation 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🔗⭐
43 RGC-VQA: An Exploration Database for Robotic-Generated Video Quality Assessment 提出RGC-VQA以解决机器人生成视频质量评估问题 egocentric ✅

⬅️ 返回 cs.CV 首页 · 🏠 返回主页