cs.CV（2025-10-07）

📊 共 33 篇论文 | 🔗 9 篇有代码

🎯 兴趣领域导航

支柱九：具身大模型 (Embodied Foundation Models) (14 🔗4) 支柱二：RL算法与架构 (RL & Architecture) (8 🔗3) 支柱三：空间感知与语义 (Perception & Semantics) (8 🔗1) 支柱一：机器人控制 (Robot Control) (2 🔗1) 支柱五：交互与反应 (Interaction & Reaction) (1)

🔬 支柱九：具身大模型 (Embodied Foundation Models) (14 篇)

#题目一句话要点标签🔗⭐
1 Detection and Measurement of Hailstones with Multimodal Large Language Models 利用多模态大语言模型检测和测量冰雹，提升恶劣天气事件评估效率。 large language model multimodal
2 Discrete Diffusion Models with MLLMs for Unified Medical Multimodal Generation 提出MeDiM，一种基于MLLM的医学多模态离散扩散模型，实现统一的医学图像和文本生成。 large language model foundation model multimodal
3 Seeing the Big Picture: Evaluating Multimodal LLMs' Ability to Interpret and Grade Handwritten Student Work 评估多模态LLM对手写学生作业的理解和评分能力 large language model multimodal
4 From Captions to Keyframes: KeyScore for Multimodal Frame Scoring and Video-Language Understanding 提出KeyScore，一种基于字幕感知的多模态关键帧评分方法，用于提升视频语言理解。 multimodal
5 Lumina-DiMOO: An Omni Diffusion Large Language Model for Multi-Modal Generation and Understanding Lumina-DiMOO：一种用于多模态生成与理解的Omni扩散大语言模型 large language model ✅
6 Multimodal Feature Prototype Learning for Interpretable and Discriminative Cancer Survival Prediction FeatProto：用于可解释和判别性癌症生存预测的多模态特征原型学习 multimodal ✅
7 BioAutoML-NAS: An End-to-End AutoML Framework for Multimodal Insect Classification via Neural Architecture Search on Large-Scale Biodiversity Data BioAutoML-NAS：基于神经架构搜索的多模态昆虫分类AutoML框架 multimodal
8 FoleyGRAM: Video-to-Audio Generation with GRAM-Aligned Multimodal Encoders FoleyGRAM：利用GRAM对齐的多模态编码器实现视频到音频的生成 multimodal
9 SD-MVSum: Script-Driven Multimodal Video Summarization Method and Datasets 提出SD-MVSum，利用跨模态注意力机制实现脚本驱动的多模态视频摘要 multimodal ✅
10 Diffusion Models for Low-Light Image Enhancement: A Multi-Perspective Taxonomy and Performance Analysis 综述论文：扩散模型在低光照图像增强中的应用、分类与性能分析 foundation model multimodal
11 Scalable deep fusion of spaceborne lidar and synthetic aperture radar for global forest structural complexity mapping 提出一种可扩展的深度融合框架，利用星载激光雷达和合成孔径雷达进行全球森林结构复杂性制图。 multimodal
12 StereoSync: Spatially-Aware Stereo Audio Generation from Video StereoSync：提出一种空间感知立体声音频生成模型，用于视频配乐。 foundation model
13 ChainMPQ: Interleaved Text-Image Reasoning Chains for Mitigating Relation Hallucinations 提出ChainMPQ，通过交错文本-图像推理链缓解关系幻觉问题 multimodal
14 Redefining Generalization in Visual Domains: A Two-Axis Framework for Fake Image Detection with FusionDetect 提出FusionDetect，融合CLIP和DINOv2特征，提升伪图像检测的泛化能力。 foundation model ✅

🔬 支柱二：RL算法与架构 (RL & Architecture) (8 篇)

#题目一句话要点标签🔗⭐
15 HOI-R1: Exploring the Potential of Multimodal Large Language Models for Human-Object Interaction Detection HOI-R1：探索多模态大语言模型在人-物交互检测中的潜力 reinforcement learning human-object interaction HOI ✅
16 Improving Chain-of-Thought Efficiency for Autoregressive Image Generation 提出ShortCoTI框架，提升自回归图像生成中思维链的效率，减少冗余计算。 reinforcement learning large language model foundation model
17 Towards Robust and Realible Multimodal Misinformation Recognition with Incomplete Modality 提出MMLNet，解决多模态信息传播中模态缺失导致的虚假信息识别鲁棒性问题。 contrastive learning multimodal ✅
18 GAZE:Governance-Aware pre-annotation for Zero-shot World Model Environments GAZE：面向零样本世界模型的治理感知预标注流水线 world model scene understanding multimodal
19 Midway Network: Learning Representations for Recognition and Motion from Latent Dynamics Midway Network：通过潜在动态学习进行识别和运动的表征学习 latent dynamics optical flow motion latent
20 When Thinking Drifts: Evidential Grounding for Robust Video Reasoning 提出Visual Evidence Reward (VER)框架，解决视频推理中思维漂移问题。 reinforcement learning multimodal chain-of-thought
21 VideoMiner: Iteratively Grounding Key Frames of Hour-Long Videos via Tree-based Group Relative Policy Optimization 提出VideoMiner，通过树状结构和强化学习优化，解决长视频关键帧提取与理解难题。 reinforcement learning spatiotemporal large language model ✅
22 Deforming Videos to Masks: Flow Matching for Referring Video Segmentation 提出FlowRVS以解决视频对象分割中的语言引导问题 flow matching

🔬 支柱三：空间感知与语义 (Perception & Semantics) (8 篇)

#题目一句话要点标签🔗⭐
23 Human3R: Everyone Everywhere All at Once Human3R：提出统一的单目视频4D人体场景重建框架，实现多人、场景和相机轨迹的实时重建。 depth estimation scene reconstruction contact-aware ✅
24 EgoNight: Towards Egocentric Vision Understanding at Night with a Challenging Benchmark EgoNight：提出首个夜间第一人称视觉理解基准，解决低光照场景下的VQA难题。 depth estimation egocentric egocentric vision
25 Flow4Agent: Long-form Video Understanding via Motion Prior from Optical Flow Flow4Agent：利用光流运动先验进行长视频理解，提升MLLM性能。 optical flow large language model multimodal
26 When and How to Cut Classical Concerts? A Multimodal Automated Video Editing Approach 提出一种多模态自动视频编辑方法，用于古典音乐会多机位录制视频的剪辑。 scene understanding multimodal
27 ArchitectHead: Continuous Level of Detail Control for 3D Gaussian Head Avatars ArchitectHead：提出首个支持连续细节层次控制的3D高斯头部头像框架 3D gaussian splatting 3DGS gaussian splatting
28 Teleportraits: Training-Free People Insertion into Any Scene Teleportraits：提出一种免训练的人物插入方法，实现任意场景下的人物合成 affordance classifier-free guidance affordance-aware
29 Human Action Recognition from Point Clouds over Time 提出一种基于点云序列和稀疏卷积网络的3D人体动作识别方法 depth estimation monocular depth
30 Dropping the D: RGB-D SLAM Without the Depth Sensor DropD-SLAM：无需深度传感器的单目RGB SLAM，达到RGB-D级别精度 metric depth

🔬 支柱一：机器人控制 (Robot Control) (2 篇)

#题目一句话要点标签🔗⭐
31 Bimanual 3D Hand Motion and Articulation Forecasting in Everyday Images 提出基于扩散模型的双手3D运动与姿态预测方法，提升日常图像中的预测精度。 bi-manual multimodal
32 HoloScene: Simulation-Ready Interactive 3D Worlds from a Single Video HoloScene：从单视频重建可交互、可仿真的3D场景 manipulation scene understanding ✅

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

#题目一句话要点标签🔗⭐
33 Text2Interact: High-Fidelity and Diverse Text-to-Two-Person Interaction Generation Text2Interact：提出高保真、多样化的文本驱动双人互动生成框架 two-person interaction spatiotemporal

⬅️ 返回 cs.CV 首页 · 🏠 返回主页