cs.CV（2025-04-24）

📊 共 31 篇论文 | 🔗 5 篇有代码

🎯 兴趣领域导航

支柱九：具身大模型 (Embodied Foundation Models) (14 🔗2) 支柱二：RL算法与架构 (RL & Architecture) (9 🔗2) 支柱三：空间感知与语义 (Perception & Semantics) (3 🔗1) 支柱六：视频提取与匹配 (Video Extraction) (2) 支柱一：机器人控制 (Robot Control) (2) 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 TimeSoccer: An End-to-End Multimodal Large Language Model for Soccer Commentary Generation 提出TimeSoccer，一个端到端多模态大语言模型，用于足球赛事解说生成。 large language model multimodal TAMP
2 TRACE: Textual Relevance Augmentation and Contextual Encoding for Multimodal Hate Detection 提出TRACE框架以解决社交媒体恶意内容检测问题 multimodal visual grounding
3 Token Sequence Compression for Efficient Multimodal Computing 提出基于聚类级别token聚合的视觉token压缩方法，提升多模态计算效率。 multimodal
4 Plasma State Monitoring and Disruption Characterization using Multimodal VAEs 提出基于多模态VAE的等离子体状态监测与破裂特征分析方法。 multimodal
5 Hierarchical and Multimodal Data for Daily Activity Understanding DARai：用于日常活动理解的分层多模态数据集，支持反事实活动分析。 multimodal
6 Fine-tune Smarter, Not Harder: Parameter-Efficient Fine-Tuning for Geospatial Foundation Models 针对地理空间基础模型，提出更智能而非更费力的参数高效微调方法。 foundation model
7 Benchmarking Multimodal Mathematical Reasoning with Explicit Visual Dependency 提出VCBENCH基准，评估LVLM在显式视觉依赖的多模态数学推理能力 multimodal ✅
8 MASR: Self-Reflective Reasoning through Multimodal Hierarchical Attention Focusing for Agent-based Video Understanding 提出MASR框架，通过多模态分层注意力自反思推理提升Agent视频理解能力 multimodal
9 FashionM3: Multimodal, Multitask, and Multiround Fashion Assistant based on Unified Vision-Language Model FashionM3：基于统一视觉-语言模型的时尚多轮多任务助手 multimodal
10 Token-Shuffle: Towards High-Resolution Image Generation with Autoregressive Models 提出Token-Shuffle，提升自回归模型在高分辨率图像生成中的效率与质量。 large language model multimodal
11 DiMeR: Disentangled Mesh Reconstruction Model DiMeR：提出解耦的网格重建模型，用于稀疏视角下的三维重建。 foundation model
12 FRAG: Frame Selection Augmented Generation for Long Video and Long Document Understanding 提出FRAG：一种帧选择增强生成框架，用于长视频和长文档理解。 multimodal ✅
13 TimeChat-Online: 80% Visual Tokens are Naturally Redundant in Streaming Videos 提出TimeChat-Online，通过差分Token丢弃解决在线视频流冗余问题。 large language model
14 VEU-Bench: Towards Comprehensive Understanding of Video Editing 提出VEU-Bench，用于评估和提升视频大语言模型在视频编辑理解方面的能力。 large language model

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

#题目一句话要点标签🔗⭐
15 Breaking the Modality Barrier: Universal Embedding Learning with Multimodal LLMs 提出UniME框架，利用多模态LLM学习通用嵌入，提升跨模态检索性能 representation learning distillation large language model
16 DPMambaIR: All-in-One Image Restoration via Degradation-Aware Prompt State Space Model DPMambaIR：基于退化感知提示状态空间模型的全能图像复原 Mamba SSM state space model
17 A Genealogy of Foundation Models in Remote Sensing 综述遥感领域Foundation Model发展，探索多传感器融合与未来方向 representation learning foundation model
18 PhysioSync: Temporal and Cross-Modal Contrastive Learning Inspired by Physiological Synchronization for EEG-Based Emotion Recognition 提出PhysioSync以解决EEG情感识别中的多模态同步问题 dreamer contrastive learning multimodal
19 DRC: Enhancing Personalized Image Generation via Disentangled Representation Composition 提出DRC，通过解耦表征组合增强个性化图像生成，缓解指导崩溃问题。 representation learning large language model multimodal
20 Mamba-Sea: A Mamba-based Framework with Global-to-Local Sequence Augmentation for Generalizable Medical Image Segmentation Mamba-Sea：基于Mamba和全局-局部序列增强的医学图像分割通用框架 Mamba state space model ✅
21 StereoMamba: Real-time and Robust Intraoperative Stereo Disparity Estimation via Long-range Spatial Dependencies StereoMamba：面向机器人辅助微创手术的实时鲁棒立体视觉视差估计 Mamba MAE
22 STCL:Curriculum learning Strategies for deep learning image steganography models 提出STCL课程学习策略，提升深度学习图像隐写模型的性能与训练效率。 curriculum learning ✅
23 Masked strategies for images with small objects 针对小目标图像，提出基于掩码策略的自监督学习方法，提升分割与分类性能。 masked autoencoder MAE

🔬 支柱三：空间感知与语义 (Perception & Semantics) (3 篇)

#题目一句话要点标签🔗⭐
24 Casual3DHDR: Deblurring High Dynamic Range 3D Gaussian Splatting from Casually Captured Videos 提出Casual3DHDR以解决高动态范围场景重建问题 3D gaussian splatting 3DGS gaussian splatting ✅
25 The Fourth Monocular Depth Estimation Challenge 第四届单目深度估计挑战赛聚焦零样本泛化，提升了自然和室内环境下的深度估计精度。 depth estimation monocular depth Depth Anything
26 Occlusion-Aware Self-Supervised Monocular Depth Estimation for Weak-Texture Endoscopic Images 提出一种遮挡感知自监督单目深度估计方法，用于弱纹理内窥镜图像。 depth estimation monocular depth

🔬 支柱六：视频提取与匹配 (Video Extraction) (2 篇)

#题目一句话要点标签🔗⭐
27 EgoCHARM: Resource-Efficient Hierarchical Activity Recognition using an Egocentric IMU Sensor EgoCHARM：利用头戴IMU传感器实现资源高效的分层活动识别 egocentric
28 Perspective-Aware Reasoning in Vision-Language Models via Mental Imagery Simulation 提出基于心智图像模拟的抽象视角变换框架，提升视觉-语言模型中的视角感知推理能力 egocentric foundation model

🔬 支柱一：机器人控制 (Robot Control) (2 篇)

#题目一句话要点标签🔗⭐
29 Towards Generalized and Training-Free Text-Guided Semantic Manipulation 提出GTF：一种通用、免训练的文本引导语义操控方法 manipulation
30 Step1X-Edit: A Practical Framework for General Image Editing Step1X-Edit：一种通用的实用图像编辑框架，性能媲美GPT-4o和Gemini2 Flash。 manipulation multimodal

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

#题目一句话要点标签🔗⭐
31 PICO: Reconstructing 3D People In Contact with Objects PICO：重建与物体接触的3D人体，解决自然图像中HOI的深度歧义问题。 human-object interaction HOI SMPL

⬅️ 返回 cs.CV 首页 · 🏠 返回主页