cs.CV（2026-06-30）

📊 共 55 篇论文 | 🔗 18 篇有代码

🎯 兴趣领域导航

支柱九：具身大模型 (Embodied Foundation Models) (18 🔗8) 支柱二：RL算法与架构 (RL & Architecture) (17 🔗5) 支柱三：空间感知与语义 (Perception & Semantics) (9 🔗2) 支柱七：动作重定向 (Motion Retargeting) (4) 支柱八：物理动画 (Physics-based Animation) (3 🔗1) 支柱一：机器人控制 (Robot Control) (3 🔗1) 支柱四：生成式动作 (Generative Motion) (1 🔗1)

🔬 支柱九：具身大模型 (Embodied Foundation Models) (18 篇)

#题目一句话要点标签🔗⭐
1 Learning to Deny: Action Denial in Multimodal Large Language Models 提出UCF101-AD基准以解决多模态模型的动作否认问题 large language model multimodal ✅
2 Attend, Transform, or Silence: Operator-Level Visual Skipping for Efficient Multimodal LLM Inference 提出操作级视觉跳过机制以提高多模态LLM推理效率 large language model multimodal
3 SimpleSearch-VL: A Simple Recipe for Multimodal Agentic Deep Search 提出SimpleSearch-VL以提升多模态智能搜索效率与可靠性 multimodal chain-of-thought
4 MS-Resampler: Multi-Scope Visual Resampling for Efficient Multimodal LLMs 提出MS-Resampler以解决多模态大语言模型的视觉重采样问题 large language model multimodal
5 Reasoning-aware Speculative Decoding for Efficient Vision-Language-Action Models in Autonomous Driving 提出推理感知的投机解码以提升自动驾驶中的视觉-语言-动作模型效率 vision-language-action VLA
6 Rethinking Foundation Model Collaboration: Enhancing Specialized Models through Proxy Task Reasoning 提出FAT框架以增强基础模型与专用模型的协作能力 foundation model multimodal
7 PruneGround: Plug-and-play Spatial Pruning for 3D Visual Grounding 提出PruneGround以解决3D视觉定位中的计算复杂性问题 language conditioned visual grounding ✅
8 Evidence Triangulation for Multimodal Fact-Checking in the Wild 提出TRENT模型以解决多模态事实核查中的数据复杂性问题 multimodal ✅
9 AA: A Multi-view Multimodal Dataset for Screen-based Gaze Estimation 提出AA数据集以解决屏幕基视线估计的多视角问题 multimodal
10 CoLT: Teaching Multi-Modal Models to Think with Chain of Latent Thoughts 提出CoLT框架以提升多模态模型的推理效率 large language model chain-of-thought ✅
11 ERA: Entropy-Guided Visual Token Pruning with Rectified Attention for Efficient MLLMs 提出ERA框架以解决多模态大语言模型的视觉令牌冗余问题 large language model multimodal ✅
12 World Narrative Model for Highly Controllable Video Generation: A Paradigm Shift from Pixel Sampling to Physical World Orchestration 提出世界叙事模型以解决视频生成可控性不足问题 foundation model multimodal
13 Learning from Failure: Inference-Time Self-Improvement for Computer-Use Agents 提出基于失败驱动的自我改进方法以提升计算机使用代理的性能 large language model multimodal
14 Beyond Single Character: Evaluating MLLMs for Sentence-Level Oracle Bone Inscription Understanding 提出S-OBI基准以解决句子级甲骨文理解问题 large language model multimodal
15 Hierarchical 3D Scene Graph Construction and Belief-based Planning for Semantic Navigation 提出层次化3D场景图构建与基于信念的规划以解决语义导航问题 foundation model
16 RESOLVE: A Multi-Resolution and Multi-Modal Dataset for Roadside Cooperative Perception 提出RESOLVE数据集以解决路边协同感知中的传感器配置问题 multimodal ✅
17 HVPNet: A Bio-Inspired Network for General Salient and Camouflaged Object Detection 提出HVPNet以解决多模态显著性与伪装物体检测问题 multimodal ✅
18 Dual Sparse Aggregation Transformer for Multispectral Object Detection 提出双稀疏聚合变换器以解决多光谱目标检测中的冗余信息问题 multimodal ✅

🔬 支柱二：RL算法与架构 (RL & Architecture) (17 篇)

#题目一句话要点标签🔗⭐
19 Breaking Failure Cascades: Step-Aware Reinforcement Learning for Medical Multimodal Reasoning 提出MRPO以解决医疗多模态推理中的失败级联问题 reinforcement learning large language model multimodal ✅
20 SAMBA: A Scatter-Guided Masked Bidirectional Mamba Foundation Model for SAR Target Recognition 提出SAMBA以解决SAR目标识别中的数据稀缺与计算复杂性问题 Mamba masked autoencoder MAE ✅
21 Robust Autonomous UAV Landing on Maritime Platforms via Multimodal Agentic AI and Active Wave Compensation 提出多模态智能体与主动波浪补偿以解决无人机海上平台安全着陆问题 reinforcement learning deep reinforcement learning DRL
22 Token-Sparse Medical Multimodal Reasoning via Dual-Stream Reinforcement Learning 提出ViToS框架以解决医疗图像稀疏信息推理问题 reinforcement learning policy learning multimodal
23 Mitigating Positional Leakage in 3D Masked Autoencoders for Robust Representation Learning 提出MPL-MAE以解决3D掩码自编码器中的位置泄漏问题 representation learning masked autoencoder MAE ✅
24 RCL-Mamba: A Dual-domain State Space Model for Measurement-oriented Image Restoration in Rotational Sparse-View Scanning Computed Laminography 提出RCL-Mamba以解决旋转稀疏视图扫描图像恢复问题 Mamba SSM state space model
25 MemLearner: Learning to Query Context memory for Video World Models 提出MemLearner以解决视频世界模型中的记忆不足问题 world model world models
26 WorldRoamBench: An Open-World Benchmark for Long-Horizon Stability of Interactive World Models 提出WorldRoamBench以解决交互世界模型的长期稳定性评估问题 world model world models
27 UniCoder: Unified Visual-to-Code Generation via Symbolic Rewards and Reference-Guided Code Optimization 提出UniCoder以解决视觉到代码生成中的奖励稀疏和探索停滞问题 reinforcement learning large language model multimodal
28 ADAPT: Attention Dynamics Alignment with Preference Tuning for Faithful MLLMs 提出ADAPT以解决多模态大语言模型的幻觉问题 DPO large language model multimodal ✅
29 One Video, One World: Turning Monocular Video into Physical 4D Scenes 提出OVOW以解决单目视频重建物理4D场景问题 world model world models embodied AI
30 Rhythm-Structured Predictive Learning for Remote Photoplethysmography 提出RhythmJEPA以解决rPPG方法的时序结构不足问题 Mamba representation learning PULSE ✅
31 Cross-Space Distillation: Teaching One-Step Students with Modern Diffusion Teachers 提出跨空间蒸馏方法以解决教师与学生潜在空间不匹配问题 distillation
32 MuSViT: A Foundation Vision Model for Sheet Music Representation 提出MuSViT以解决乐谱表示的基础模型缺失问题 masked autoencoder foundation model
33 Unsupervised Data-Efficient Cross-Modal Retrieval with Global-Neighborhood Alignment Hashing 提出全球邻域对齐哈希以解决无监督跨模态检索问题 contrastive learning foundation model
34 PRISM: Latent Composition Consistency for Single-Image Reflection Removal 提出PRISM以解决单幅图像反射去除问题 flow matching contrastive learning
35 CLIMB: Centroid-Based Hierarchical Memory for Online Continual Self-Supervised Learning 提出CLIMB以解决在线持续自监督学习中的记忆管理问题 contrastive learning distillation

🔬 支柱三：空间感知与语义 (Perception & Semantics) (9 篇)

#题目一句话要点标签🔗⭐
36 AugSplat: Radiance Field-Informed Gaussian Splatting for Sparse-View Settings 提出AugSplat以解决稀疏视图下的高质量3D重建问题 gaussian splatting splatting NeRF
37 PointSplat: Compact Gaussian Splatting via Human-Centric Prediction 提出PointSplat以解决3D人类表示的冗余问题 gaussian splatting splatting
38 DrivingDepth: Sparse-Prompted Pixel-wise Scale Correction for Driving Depth Estimation 提出DrivingDepth以解决自动驾驶深度估计中的几何尺度冲突问题 depth estimation geometric consistency foundation model
39 CasaMaestro: Multi-View Panoramas for House-Scale 3D Reconstruction 提出CasaMaestro以解决住宅空间3D重建效率问题 metric depth 3D reconstruction embodied AI
40 WarpHammer: Densifying Scene Warps with 3D Object Priors for Extreme View Synthesis 提出WarpHammer以解决极端视角合成中的稀疏变形问题 3D reconstruction sam 3D SAM 3D
41 Learning Video Dynamics with Predictive Differentiable Rendering 提出预测可微渲染以解决视频预测中的细节缺失问题 3D gaussian splatting 3D reconstruction gaussian splatting
42 Think While You Map: Asynchronous Vision-Language Agents for Incremental 3D Scene Graphs 提出异步视觉语言代理以解决增量3D场景图构建问题 open-vocabulary open vocabulary visual grounding ✅
43 AnyMatch: Supercharging Universal Multi-Modal Image Matching with Large-Scale Single-View Images 提出AnyMatch以解决多模态图像匹配数据稀缺问题 depth estimation monocular depth geometric consistency
44 LiteMatch: Lightweight Zero-Shot Stereo Matching via Cost Volume Stabilization 提出LiteMatch以解决轻量级零-shot立体匹配问题 scene flow ✅

🔬 支柱七：动作重定向 (Motion Retargeting) (4 篇)

#题目一句话要点标签🔗⭐
45 WildProp: Visual Estimation of Wildlife Body Proportions at Scale 提出WildProp以解决野生动物体型比例估计问题 geometric consistency foundation model
46 LUNA: Learning Universal 3D Human Animation Beyond Skinning 提出LUNA以解决单目图像生成3D人类动画的局限性 human motion
47 Accelerating Merge with Motion Vector Difference via Filter Difference Analysis for VVenC 提出快速MMVD算法以降低编码器计算负担 motion prediction
48 Estimating Velocity of Spheres from Rolling-Shutter Image(s) 利用滚动快门图像估计球体的速度 geometric consistency

🔬 支柱八：物理动画 (Physics-based Animation) (3 篇)

#题目一句话要点标签🔗⭐
49 Temporal Preservation over Processing: Diagnosing and Designing Spatiotemporal Single-Stage Video Detectors 提出TemporalLens与YOLO-3D以解决视频检测中的时间推理问题 spatiotemporal
50 FLORA: A deep learning approach to predict forest attributes from heterogeneous LiDAR data 提出FLORA以解决异构LiDAR数据下森林属性预测问题 spatiotemporal
51 Temperature Field Reconstruction of Tungsten Monoblock Divertor on EAST using Physics-aware Neural Operator Transformer 提出物理感知神经算子变换器以解决东亚聚变装置温度场重建问题 spatiotemporal ✅

🔬 支柱一：机器人控制 (Robot Control) (3 篇)

#题目一句话要点标签🔗⭐
52 Editing Everything Everywhere All at Once 提出MICE以解决多实例图像编辑中的语义干扰问题 manipulation multimodal
53 Look But Don't Touch with Sparse Autoencoders for Unlearning in Diffusion Models 提出稀疏自编码器以解决扩散模型中的概念干预问题 manipulation
54 MV-GEL: Language-Driven Multi-View Geometric Entity Localization on Meshes 提出MV-GEL框架以解决3D几何实体定位问题 manipulation ✅

🔬 支柱四：生成式动作 (Generative Motion) (1 篇)

#题目一句话要点标签🔗⭐
55 SpheRoPE: Zero-Shot Optimization-Free 360 Panorama Generation with Spherical RoPE 提出SpheRoPE以解决360全景图生成中的优化与训练问题 classifier-free guidance ✅

⬅️ 返回 cs.CV 首页 · 🏠 返回主页