cs.CV（2026-05-01）

📊 共 24 篇论文 | 🔗 7 篇有代码

🎯 兴趣领域导航

支柱九：具身大模型 (Embodied Foundation Models) (9 🔗3) 支柱三：空间感知与语义 (Perception & Semantics) (5 🔗1) 支柱二：RL算法与架构 (RL & Architecture) (4 🔗2) 支柱四：生成式动作 (Generative Motion) (2) 支柱一：机器人控制 (Robot Control) (2) 支柱八：物理动画 (Physics-based Animation) (2 🔗1)

🔬 支柱九：具身大模型 (Embodied Foundation Models) (9 篇)

#题目一句话要点标签🔗⭐
1 LIMSSR: LLM-Driven Sequence-to-Score Reasoning under Training-Time Incomplete Multimodal Observations 提出LIMSSR框架，解决训练时多模态数据不完整情况下的序列到评分推理问题 large language model multimodal ✅
2 UniVidX: A Unified Multimodal Framework for Versatile Video Generation via Diffusion Priors UniVidX：基于扩散先验的统一多模态视频生成框架 multimodal ✅
3 Let ViT Speak: Generative Language-Image Pre-training GenLIP：面向多模态大语言模型的生成式语言-图像预训练框架 large language model multimodal
4 High-Speed Vision Improves Zero-Shot Semantic Understanding of Human Actions 高帧率视频提升零样本语义理解在高速人类动作识别中的性能 large language model
5 Persistent Visual Memory: Sustaining Perception for Deep Generation in LVLMs 提出持久视觉记忆（PVM）模块，解决LVLM深度生成中的视觉信号稀释问题。 multimodal
6 Make Your LVLM KV Cache More Lightweight LightKV：通过提示引导的跨模态压缩，减少LVLM中KV缓存的内存占用。 large language model
7 BlenderRAG: High-Fidelity 3D Object Generation via Retrieval-Augmented Code Synthesis BlenderRAG：通过检索增强的代码合成实现高保真3D对象生成 multimodal ✅
8 MMAudio-LABEL: Audio Event Labeling via Audio Generation for Silent Video 提出MMAudio-LABEL框架，通过联合生成音频和事件标签，提升无声视频的音频事件标注性能。 multimodal
9 Scaling Video Understanding via Compact Latent Multi-Agent Collaboration 提出MACF：通过紧凑潜在多智能体协作实现可扩展的视频理解 large language model

🔬 支柱三：空间感知与语义 (Perception & Semantics) (5 篇)

#题目一句话要点标签🔗⭐
10 2D-SuGaR: Surface-Aware Gaussian Splatting for Geometrically Accurate Mesh Reconstruction 提出2D-SuGaR，利用单目深度和法向量先验提升2D高斯溅射的几何重建精度。 monocular depth 3D gaussian splatting 3DGS
11 GOR-IS: 3D Gaussian Object Removal in the Intrinsic Space 提出GOR-IS，在内参空间实现3D高斯模型的物体移除与光照一致性修复。 3D gaussian splatting 3DGS gaussian splatting ✅
12 Foundation AI Models for Aerosol Optical Depth Estimation from PACE Satellite Data 提出ViTCG，利用Transformer和通道分组进行气溶胶光学厚度估计，显著降低误差。 depth estimation foundation model
13 Modeling Subjective Urban Perception with Human Gaze 提出Place Pulse-Gaze数据集和Gaze-Guided框架，利用人类注视建模主观城市感知。 scene understanding PULSE multimodal
14 Pose-Aware Diffusion for 3D Generation 提出姿态感知扩散模型PAD，用于生成姿态对齐的3D物体，解决空间错位和变换歧义问题。 monocular depth scene reconstruction

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

#题目一句话要点标签🔗⭐
15 Beyond Heuristics: Learnable Density Control for 3D Gaussian Splatting LeGS：基于强化学习的可学习密度控制，提升3D高斯溅射渲染质量 reinforcement learning 3D gaussian splatting 3DGS ✅
16 Learning from Compressed CT: Feature Attention Style Transfer and Structured Factorized Projections for Resource-Efficient Medical Image Analysis 提出FAST和SFP，用于压缩CT图像的资源高效医学图像分析 contrastive learning distillation spatiotemporal
17 Posterior Augmented Flow Matching 提出后验增强Flow Matching，解决高维图像生成中Flow Collapse问题 flow matching ✅
18 Online Self-Calibration Against Hallucination in Vision-Language Models 提出OSCAR框架，在线自校准视觉-语言模型中的幻觉问题 direct preference optimization multimodal

🔬 支柱四：生成式动作 (Generative Motion) (2 篇)

#题目一句话要点标签🔗⭐
19 PhysiGen: Integrating Collision-Aware Physical Constraints for High-Fidelity Human-Human Interaction Generation PhysiGen：集成碰撞感知物理约束，实现高保真的人人交互生成 motion synthesis penetration multi-person interaction
20 Robust Fusion of Object-Level V2X for Learned 3D Object Detection 提出噪声感知训练策略，提升V2X融合3D目标检测在噪声环境下的鲁棒性 penetration

🔬 支柱一：机器人控制 (Robot Control) (2 篇)

#题目一句话要点标签🔗⭐
21 Colorful-Noise: Training-Free Low-Frequency Noise Manipulation for Color-Based Conditional Image Generation 提出Colorful-Noise，通过无训练的低频噪声操控实现彩色条件图像生成。 manipulation
22 From Local to Global to Mechanistic: An iERF-Centered Unified Framework for Interpreting Vision Models 提出iERF中心统一框架，实现视觉模型局部、全局和机制可解释性 manipulation

🔬 支柱八：物理动画 (Physics-based Animation) (2 篇)

#题目一句话要点标签🔗⭐
23 MMAudioReverbs: Video-Guided Acoustic Modeling for Dereverberation and Room Impulse Response Estimation 提出MMAudioReverbs，利用视频引导的声学建模进行解混响和房间脉冲响应估计 PULSE
24 CMTA: Leveraging Cross-Modal Temporal Artifacts for Generalizable AI-Generated Video Detection CMTA：利用跨模态时间伪影实现通用AI生成视频检测 spatiotemporal ✅

⬅️ 返回 cs.CV 首页 · 🏠 返回主页