cs.CV（2025-04-23）

📊 共 27 篇论文 | 🔗 9 篇有代码

🎯 兴趣领域导航

支柱二：RL算法与架构 (RL & Architecture) (9 🔗5) 支柱三：空间感知与语义 (Perception & Semantics) (6 🔗1) 支柱九：具身大模型 (Embodied Foundation Models) (5 🔗1) 支柱八：物理动画 (Physics-based Animation) (4 🔗1) 支柱一：机器人控制 (Robot Control) (1 🔗1) 支柱五：交互与反应 (Interaction & Reaction) (1) 支柱六：视频提取与匹配 (Video Extraction) (1)

🔬 支柱二：RL算法与架构 (RL & Architecture) (9 篇)

#题目一句话要点标签🔗⭐
1 PMG: Progressive Motion Generation via Sparse Anchor Postures Curriculum Learning 提出ProMoGen框架，通过稀疏锚点姿势和课程学习实现可控、高保真的人体动作生成。 curriculum learning motion synthesis motion generation
2 Skywork R1V2: Multimodal Hybrid Reinforcement Learning for Reasoning Skywork R1V2提出混合强化学习框架，提升多模态推理能力并缓解视觉幻觉问题。 reinforcement learning multimodal ✅
3 Federated EndoViT: Pretraining Vision Transformers via Federated Learning on Endoscopic Image Collections 提出基于联邦学习的内窥镜Vision Transformer预训练方法，解决数据共享限制。 masked autoencoder MAE spatiotemporal
4 Decoupled Global-Local Alignment for Improving Compositional Understanding 提出解耦全局-局部对齐框架DeGLA，提升CLIP模型组合理解能力并保持通用性。 contrastive learning distillation large language model ✅
5 A Diff-Attention Aware State Space Fusion Model for Remote Sensing Classification 提出DAS2F-Model，通过差异注意力机制和状态空间模型融合多光谱与全色遥感影像，提升分类精度。 Mamba state space model multimodal ✅
6 Revisiting Radar Camera Alignment by Contrastive Learning for 3D Object Detection 提出RCAlign模型，通过对比学习实现雷达相机特征对齐，提升3D目标检测精度。 contrastive learning distillation
7 JEPA for RL: Investigating Joint-Embedding Predictive Architectures for Reinforcement Learning 提出基于JEPA的强化学习框架，解决图像强化学习中的表征学习问题 reinforcement learning
8 Rethinking Generalizable Infrared Small Target Detection: A Real-scene Benchmark and Cross-view Representation Learning 提出基于跨视角表征学习的红外小目标检测框架，提升泛化能力 representation learning ✅
9 CLPSTNet: A Progressive Multi-Scale Convolutional Steganography Model Integrating Curriculum Learning 提出CLPSTNet以解决图像隐写中的隐蔽性与安全性问题 curriculum learning ✅

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

#题目一句话要点标签🔗⭐
10 TraveLLaMA: A Multimodal Travel Assistant with Large-Scale Dataset and Structured Reasoning TraveLLaMA：基于大规模数据集和结构化推理的多模态旅行助手 scene understanding multimodal chain-of-thought
11 Gaussian Splatting is an Effective Data Generator for 3D Object Detection 利用高斯溅射进行数据增强，提升自动驾驶3D目标检测性能 gaussian splatting splatting
12 Visibility-Uncertainty-guided 3D Gaussian Inpainting via Scene Conceptional Learning 提出VISTA框架，通过可见性不确定性引导和场景概念学习实现高质量3D高斯补全。 3D gaussian splatting 3DGS gaussian splatting
13 ToF-Splatting: Dense SLAM using Sparse Time-of-Flight Depth and Multi-Frame Integration ToF-Splatting：利用稀疏ToF深度和多帧融合的稠密SLAM 3D gaussian splatting gaussian splatting splatting
14 Dual-Camera All-in-Focus Neural Radiance Fields 提出DC-NeRF，利用双摄像头合成全聚焦神经辐射场，无需手动重聚焦。 NeRF neural radiance field
15 SaENeRF: Suppressing Artifacts in Event-based Neural Radiance Fields SaENeRF：提出自监督框架，抑制基于事件的神经辐射场重建伪影 NeRF neural radiance field ✅

🔬 支柱九：具身大模型 (Embodied Foundation Models) (5 篇)

#题目一句话要点标签🔗⭐
16 DyMU: Dynamic Merging and Virtual Unmerging for Efficient VLMs DyMU：动态合并与虚拟解合并，提升视觉语言模型效率且无需训练。 large language model ✅
17 Unveiling the Lack of LVLM Robustness to Fundamental Visual Variations: Why and Path Forward 揭示LVLM对基本视觉变化的鲁棒性不足，并提出改进方向 multimodal
18 VideoVista-CulturalLingo: 360$^\circ$ Horizons-Bridging Cultures, Languages, and Domains in Video Comprehension 提出VideoVista-CulturalLingo，弥合文化、语言和领域差异的视频理解评估基准。 multimodal
19 Facial Foundational Model Advances Early Warning of Coronary Artery Disease from Live Videos with DigitalShadow DigitalShadow：利用面部基础模型从实时视频中早期预警冠状动脉疾病 foundation model
20 Think Hierarchically, Act Dynamically: Hierarchical Multi-modal Fusion and Reasoning for Vision-and-Language Navigation 提出多层融合推理架构MFRA，提升视觉语言导航任务的决策精度。 VLN

🔬 支柱八：物理动画 (Physics-based Animation) (4 篇)

#题目一句话要点标签🔗⭐
21 A multi-scale vision transformer-based multimodal GeoAI model for mapping Arctic permafrost thaw 提出基于多尺度Vision Transformer的多模态GeoAI模型，用于精确绘制北极多年冻土融化滑坡 spatiotemporal multimodal
22 4D Multimodal Co-attention Fusion Network with Latent Contrastive Alignment for Alzheimer's Diagnosis 提出M2M-AlignNet，通过多模态对齐和协同注意力融合sMRI和fMRI，用于阿尔茨海默病诊断。 spatiotemporal multimodal
23 Direct Video-Based Spatiotemporal Deep Learning for Cattle Lameness Detection 提出基于视频时空深度学习的牛跛足检测方法，无需姿态估计预处理。 spatiotemporal
24 BadVideo: Stealthy Backdoor Attack against Text-to-Video Generation 提出BadVideo，针对文本生成视频模型的隐蔽后门攻击框架 spatiotemporal ✅

🔬 支柱一：机器人控制 (Robot Control) (1 篇)

#题目一句话要点标签🔗⭐
25 PPS-Ctrl: Controllable Sim-to-Real Translation for Colonoscopy Depth Estimation PPS-Ctrl：用于结肠镜深度估计的可控Sim-to-Real图像转换 sim-to-real depth estimation ✅

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

#题目一句话要点标签🔗⭐
26 A Time Series Dataset of NIR Spectra and RGB and NIR-HSI Images of the Barley Germination Process 发布大麦发芽过程的近红外光谱、RGB和近红外高光谱图像时间序列数据集 HSI

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

#题目一句话要点标签🔗⭐
27 PRaDA: Projective Radial Distortion Averaging 提出PRaDA，在射影空间中实现鲁棒的径向畸变相机自动标定 feature matching

⬅️ 返回 cs.CV 首页 · 🏠 返回主页