cs.CV（2026-07-05）

📊 共 99 篇论文

🎯 兴趣领域导航

支柱九：具身大模型 (Embodied Foundation Models) (27) 支柱三：空间感知与语义 (Perception & Semantics) (26) 支柱二：RL算法与架构 (RL & Architecture) (25) 支柱一：机器人控制 (Robot Control) (8) 支柱六：视频提取与匹配 (Video Extraction) (5) 支柱四：生成式动作 (Generative Motion) (4) 支柱五：交互与反应 (Interaction & Reaction) (2) 支柱八：物理动画 (Physics-based Animation) (1) 支柱七：动作重定向 (Motion Retargeting) (1)

🔬 支柱九：具身大模型 (Embodied Foundation Models) (27 篇)

#题目一句话要点标签🔗⭐
1 LLM-Empowered Multimodal Fusion Framework for Autonomous Driving: Semantic Enhancement and Channel-Adaptive Design 提出LM-SCIP框架以解决动态输入质量对自动驾驶感知的挑战 large language model multimodal
2 SPAR: Semantic-Pixel Self-Alignment and Adaptive Routing for Unified Multimodal Models 提出SPAR框架以解决多模态模型的视觉生成问题 large language model multimodal
3 MMBench-Live: A Continuously Evolving Benchmark for Multimodal Models 提出MMBench-Live以解决多模态基准评估的静态问题 multimodal
4 Descriptor: LYNRED Mobility Dataset Multimodal Detection Subset (LYNRED-MDS) 提出LYNRED-MDS数据集以解决低能见度条件下的碰撞预测问题 multimodal
5 Learning Spectral and Polarimetric Clues for One-to-Multimodal Novel View Synthesis 提出SPoILeR以解决多模态新视图合成问题 multimodal
6 Towards Cellular-Scale Interpretability in Pathology Foundation Models for Biomarker Assessment 提出Hireca与CytoMap以解决病理生物标志物评估的可解释性问题 foundation model
7 BRIGHT: A Collaborative Generalist-Specialist Foundation Model for Breast Pathology 提出BRIGHT以解决乳腺病理领域的特定任务挑战 foundation model
8 Rapidly deploying on-device eye tracking by distilling visual foundation models 提出DistillGaze以解决快速部署高精度眼动追踪问题 foundation model
9 Temporal and Cross-Modal Alignment for Enhanced Audiovisual Video Captioning 提出TCA-Captioner以解决视听视频字幕中的时间与跨模态对齐问题 large language model multimodal
10 ReQuest: Rethinking-based Question-Aware Frame Selection for Long-Form Video QA 提出ReQuest以解决长视频问答中的关键帧选择问题 large language model multimodal
11 EAGLE-360: Embodied Active Global-to-Local Exploration in 360$^\circ$ 提出EAGLE-360以解决360°全景环境中的主动视觉搜索问题 large language model multimodal
12 SEPS: Semantic-enhanced Patch Slimming Framework for fine-grained cross-modal alignment 提出SEPS框架以解决细粒度跨模态对齐问题 large language model multimodal
13 NarrativeTrack: Evaluating Entity-Centric Reasoning for Narrative Understanding 提出NarrativeTrack以解决多模态大语言模型叙事理解问题 large language model multimodal
14 Omni-DuplexEval: Evaluating Real-time Duplex Omni-modal Interaction 提出Omni-DuplexEval以解决实时双向多模态交互评估问题 large language model multimodal
15 LASER: A Corrective Lens for LVLMs via Visual Attention Preservation and Sink Suppression 提出LASER以解决长视角解码中的视觉遗忘问题 visual grounding
16 SAB-LVLM: Significance-Aware Binarization for Large Vision-Language Models 提出SAB-LVLM以解决大规模视觉语言模型的二值化挑战 multimodal
17 ESC: Emotional Self-Correction for Reliable Vision-Language Models 提出情感自我修正框架以提升视觉语言模型的可靠性 multimodal
18 AdaCount: Training-Free Similarity-Guided Spatial and Feature Adaptation for Zero-Shot Object Counting 提出AdaCount以解决零-shot目标计数中的实例分离问题 foundation model
19 Combating Textual Noise and Redundancy: Entropy-Aware Dense Visual Token Pruning 提出熵感知密集视觉标记修剪以解决文本噪声问题 multimodal
20 Alignment Is All You Need For X-to-4D Generation 提出Align4D以解决X到4D生成中的多模态挑战 multimodal
21 PPTArena: A Benchmark for PowerPoint Editing 提出PPTArena基准以解决PowerPoint编辑挑战 instruction following
22 TimeChat-Captioner: Scripting Multi-Scene Videos with Time-Aware and Structural Audio-Visual Captions 提出Omni Dense Captioning以解决多场景视频描述问题 TAMP
23 WikiCLIP: An Efficient Contrastive Baseline for Open-domain Visual Entity Recognition 提出WikiCLIP以解决开放域视觉实体识别的高计算成本问题 large language model
24 EPOFusion: Exposure aware Progressive Optimization Method for Infrared and Visible Image Fusion 提出EPOFusion以解决可见光图像过曝问题 multimodal
25 Beyond Absolute Scores: Relative Edit-induced Difference for Generalizable Image Aesthetic Assessment 提出RED-Aes框架以解决图像美学评估中的绝对评分局限问题 chain-of-thought
26 On Test-Time Scaling for Vision-Language Models 提出测试时间缩放方法以提升视觉语言模型性能 large language model
27 SpiralFovea: Input-Adaptive Foveated Tokenization as a Third Lever of Resource-Adaptive Inference 提出SpiralFovea以实现输入自适应的视觉令牌化 foundation model

🔬 支柱三：空间感知与语义 (Perception & Semantics) (26 篇)

#题目一句话要点标签🔗⭐
28 Consistent Scene Understanding in 3D Gaussian Splatting via Multi-Cue Mask Refinement 提出多线索掩膜优化框架以解决3D高斯点云一致性问题 3D gaussian splatting 3DGS gaussian splatting
29 Structure-Aware Gaussian Splatting for Large-Scale Scene Reconstruction 提出结构感知高斯点云以解决大规模场景重建问题 3D gaussian splatting gaussian splatting splatting
30 AnchorSplat: Fast and Structure Consistent Detail Synthesis for Gaussian Splatting 提出AnchorSplat以解决3D高保真渲染细节合成问题 3D gaussian splatting 3DGS gaussian splatting
31 MVFusion-GS: Motion-Variance Guided Temporal Attention for High-Quality Dynamic Gaussian Splatting 提出MVFusion-GS以解决动态场景重建中的运动感知不足问题 3D gaussian splatting 3DGS gaussian splatting
32 VisionAId: An Offline-First Multimodal Android Assistant for People with Visual Impairment, Featuring Personalized Object Retrieval 提出VisionAId以解决视障人士日常生活中的障碍问题 depth estimation monocular depth metric depth
33 The Turning Point of 3D Plant Phenotyping: 3D Foundation Models Enable Minute-to-Second Cross-Crop Reconstruction and Beyond 提出3D基础模型以解决3D植物表型高通量重建问题 3D gaussian splatting 3D reconstruction gaussian splatting
34 Mind the Gap: Standard 3DGS Evaluation Primarily Measures Near-Trajectory Interpolation 提出公平匹配计数协议以解决3D高斯点云评估中的插值与外推问题 3D gaussian splatting 3DGS gaussian splatting
35 COVTrack++: Learning Open-Vocabulary Multi-Object Tracking from Continuous Videos via a Synergistic Paradigm 提出COVTrack++以解决开放词汇多目标跟踪问题 open-vocabulary open vocabulary spatial relationship
36 X-Splat: Gaussian Splatting for 3D CBCT Generation from Single Panoramic Radiograph 提出X-Splat框架以解决单幅全景影像生成3D CBCT问题 gaussian splatting splatting NeRF
37 FoundDP: Revisiting Weak Disparity Observability in Dual-Pixel Depth Estimation 提出FoundDP以解决双像素深度估计中的弱视差可观测性问题 depth estimation monocular depth metric depth
38 Fisheye3R: Adapting Unified 3D Feed-Forward Foundation Models to Fisheye Lenses 提出Fisheye3R以解决鱼眼镜头图像的3D重建问题 3D reconstruction VGGT foundation model
39 ProCal: Inference-Time Proposal Calibration for Open-Vocabulary Object Detection 提出ProCal以解决开放词汇目标检测中的定位问题 open-vocabulary open vocabulary
40 MiraGe: Editable 2D Images using Gaussian Splatting 提出MiraGe以实现可编辑的2D图像 gaussian splatting splatting
41 GADA: Geometry-Aware Deformable Aggregation for Image-Based Gaussian Splatting 提出GADA以解决高频细节失真问题 gaussian splatting splatting
42 Towards Robust Driving Perception: A Flexible Scale-Driven Family for Self-Supervised Monocular Depth Estimation 提出FlexDepth以解决复杂驾驶环境中的单目深度估计问题 depth estimation monocular depth
43 Diversity-aware View Partitioning for Scalable VGGT 提出多样性感知视图分区以解决VGGT扩展性问题 3D reconstruction VGGT spatial relationship
44 Argus: Metric Panoramic 3D Reconstruction for Indoor Scenes 提出Argus以解决室内场景的度量全景3D重建问题 depth estimation 3D reconstruction geometric consistency
45 InvSplat: Inverse Feed-Forward Scene Splatting 提出Inverse Feed-Forward Scene Splatting以解决逆渲染中的多视图不一致问题 3D reconstruction splatting
46 Resonant Brane Splatting for Arbitrary-Scale Super-Resolution 提出共振膜体喷溅以解决任意尺度超分辨率问题 gaussian splatting splatting
47 ICDepth: Taming Video Diffusion Models for Video Depth Estimation via In-Context Conditioning 提出ICDepth以解决视频深度估计中的时序一致性问题 depth estimation
48 Search-based Testing of Vision Language Models for In-Car Scene Understanding 提出ISU-Test以解决车内场景理解测试问题 scene understanding
49 Gaussians on Fire: High-Frequency Reconstruction of Flames 提出基于高斯模型的动态火焰三维重建方法 monocular depth optical flow spatiotemporal
50 QWERTY: Training-Free Motion Control via Query-Warped Video Diffusion Transformers 提出QWERTY框架以实现无训练的运动控制 optical flow latent optimization
51 DL-VINS-Factory: A Modular Framework for Learned Visual Front-Ends in Visual-Inertial SLAM 提出DL-VINS-Factory以解决视觉惯性SLAM中的特征提取问题 optical flow
52 PointDiT: Pixel-Space Diffusion for Monocular Geometry Estimation 提出PointDiT以简化单图像3D重建方法 3D reconstruction
53 DarkVGGT: Seeing Through Darkness Using Thermal Geometry without Daylight Tax 提出DarkVGGT以解决低光环境下的3D重建问题 3D reconstruction

🔬 支柱二：RL算法与架构 (RL & Architecture) (25 篇)

#题目一句话要点标签🔗⭐
54 Drive-JEPA: Video JEPA Meets Multimodal Trajectory Distillation for End-to-End Driving 提出Drive-JEPA以解决自主驾驶中的多模态行为学习问题 world model world models JEPA
55 Bridging 3D Gaussians and Semantic Occupancy for Comprehensive Open-Vocabulary Scene Understanding from Unposed Images 提出COVScene以解决无姿态图像的全面3D场景理解问题 distillation scene understanding open-vocabulary
56 Geometric Foundation Model Distillation for Efficient Lunar 3D Reconstruction 通过几何基础模型蒸馏提升月球3D重建效率 distillation 3D reconstruction foundation model
57 Teaching Vision-Language-Action Models What to See and Where to Look 提出DriveTeach-VLA以解决VLA模型空间依赖性不足问题 distillation vision-language-action VLA
58 Wiki-R1: Incentivizing Multimodal Reasoning for Knowledge-based VQA via Data and Sampling Curriculum 提出Wiki-R1以解决知识基础视觉问答中的多模态推理问题 reinforcement learning large language model multimodal
59 Reliability-Prioritized Fine-Grained Generation in Multimodal Large 提出GranFact以解决多模态大语言模型的可靠性挑战 direct preference optimization large language model multimodal
60 Multimodal Fusion for Fine-Grained Classification of Breast Fibroadenoma and Phyllodes Tumors 提出多模态融合方法以解决乳腺肿瘤分类问题 representation learning multimodal
61 RGB-Pointmap Pretraining for Unified 3D Scene Understanding 提出UniScene3D以解决统一3D场景理解问题 representation learning scene understanding foundation model
62 PWM-ArtGen: Part World Model for Articulated Object Generation 提出PWM-ArtGen以解决关节物体生成中的运动结构预测问题 world model world models
63 DetailAnywhere: Fashion Detail Generation via Cross-Modal Feature Alignment Distillation 提出DetailAnywhere以解决时尚细节生成问题 reinforcement learning distillation multimodal
64 MemoBench: Benchmarking World Modeling in Dynamically Changing Environments 提出MemoBench以解决动态环境中物体记忆一致性问题 world model world models
65 WorldOdysseyBench: An Open-World Benchmark for Long-Horizon Stability of Interactive World Models 提出WorldOdysseyBench以评估交互世界模型的长期稳定性 world model world models
66 SCLARO: A Dataset for Grounded Scenario-Level Scene Understanding and ScenarioCLIP for Benchmarking 提出SCLARO数据集与ScenarioCLIP模型以解决场景理解问题 distillation scene understanding
67 Multi-Resolution Flow Matching: Training-Free Diffusion Acceleration via Staged Sampling 提出MrFlow以解决文本到图像扩散加速问题 flow matching distillation
68 TempAct: Advancing Temporal Plausibility in Autoregressive Video Generation via Planner-Executor RL 提出TempAct以解决自回归视频生成中的时间一致性问题 reinforcement learning distillation instruction following
69 RTE-FM-Dehazer: Radiative Transfer Equation Inspired Flow Matching for Real-World Image Dehazing 提出RTE-FM-Dehazer以解决图像去雾中的残留雾霾问题 flow matching
70 SFKD: Spatial--Frequency Joint-Aware Heterogeneous Knowledge Distillation via Multi-Level Wavelet Spectral Interaction 提出SFKD框架以解决异构模型间知识蒸馏问题 distillation
71 Efficient PEFT Methods with Adaptive Checkpointing for Vision Models and VLMs on Resource Constrained Consumer-GPUs 提出自适应检查点的高效PEFT方法以解决资源受限GPU的视觉模型调优问题 Mamba foundation model
72 FlowCIR: Semantic Transport via Flow Matching for Zero-Shot Composed Image Retrieval 提出FlowCIR以解决零-shot组合图像检索中的语义传输问题 flow matching
73 Dual-Selective Network for Domain-Incremental Change Detection 提出双选择网络以解决领域增量变化检测问题 Mamba state space model distillation
74 Seek to Segment: Active Perception for Panoramic Referring Segmentation 提出PanoSeeker以解决全景指向分割中的主动感知问题 reinforcement learning embodied AI
75 Occlusion-Aware 3D Hand-Object Pose Estimation with Masked AutoEncoders 提出基于掩码自编码器的手-物体姿态估计方法以解决遮挡问题 masked autoencoder
76 SGMatch: Semantic-Guided Non-Rigid Shape Matching with Flow Regularization 提出SGMatch以解决非刚性3D形状匹配问题 flow matching foundation model
77 Understanding Geometric Representations in Self-Supervised Vision Transformers via Subspace Intervention 提出受控子空间干预框架以研究自监督视觉变换器的几何表示 masked autoencoder MAE
78 WorldDirector: Building Controllable World Simulators with Persistent Dynamic Memory 提出WorldDirector以解决动态物体记忆与视角控制问题 world model world models

🔬 支柱一：机器人控制 (Robot Control) (8 篇)

#题目一句话要点标签🔗⭐
79 Motion-Focused Latent Action Enables Cross-Embodiment VLA Training from Human EgoVideos 提出基于潜在动作的框架以解决无标签人类视频训练问题 manipulation VQ-VAE egocentric
80 ComplexMimic: Human-Scene Interaction Imitation in Complex 3D Environments 提出ComplexMimic以解决复杂环境下人机交互模仿问题 motion tracking imitation learning distillation
81 VLAFlow: A Unified Training Framework for Vision-Language-Action Models via Co-training and Future Latent Alignment 提出VLAFlow框架以统一比较视觉-语言-动作模型的训练效果 manipulation flow matching vision-language-action
82 From Forgeries to Foundation Models: A Systematic Survey of Identity Document Attack and Detection 提出统一模型以解决身份文件伪造与检测问题 manipulation foundation model multimodal
83 Online Segment 3D Gaussians via Launching Virtual Drones 提出SAGO框架以解决3D高斯在线分割问题 manipulation distillation 3D gaussian splatting
84 TabletopGen: Tabletop Scene Generation and Interactive Simulation for Robotic Manipulation 提出TabletopGen以解决桌面场景生成与交互模拟问题 manipulation sim-to-real multimodal
85 Spintronics for image recognition: performance benchmarking via data-driven simulations 基于自旋电子学的图像识别方法及性能基准测试 manipulation
86 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

🔬 支柱六：视频提取与匹配 (Video Extraction) (5 篇)

#题目一句话要点标签🔗⭐
87 Graph it first! Enabling Reasoning on Long-form Egocentric Videos through Scene Graphs 提出Egocentric Scene Graphs以解决长视频理解问题 egocentric large language model
88 How Much Future Helps? A Controlled Study of Future-Privileged Supervision for Causal Egocentric Gaze Estimation 提出未来特权监督以提升因果自我中心注视估计的准确性 egocentric Ego4D
89 Multi-THuMBS: Multi-person Tracking of 3D Human Meshes Beyond Video Shots 提出Multi-THuMBS以解决多人物体3D网格跟踪中的镜头变化问题 human mesh recovery motion reconstruction
90 HandsOnWorld: Unconstrained Egocentric Video Generation with Camera-Disentangled Hand Control 提出HandsOnWorld以解决无标记手控视频生成问题 egocentric
91 Learning to Evolve Scenes: Reasoning about Human Activities with Scene Graphs 提出SG-Ego和GLEN以解决人类活动理解问题 Ego4D embodied AI

🔬 支柱四：生成式动作 (Generative Motion) (4 篇)

#题目一句话要点标签🔗⭐
92 Language-Guided Transformer Tokenizer for Human Motion Generation 提出语言引导的Transformer分词器以解决人类动作生成问题 motion generation human motion human motion generation
93 Training-free Controllable Human Motion Generation under Heterogeneous Constraints 提出Motion-Inference-as-Control框架以解决异构约束下的无训练人类动作生成问题 motion generation human motion human motion generation
94 JointHOI: Jointly Generating Contact Maps Enhances Hand Object Interaction Generation 提出JointHOI以解决手物体交互生成中的接触问题 physically plausible penetration HOI
95 PatternGSL: A Structured Specification Language for Template-Free and Simulation-Ready 3D Garments 提出PatternGSL以解决无模板3D服装重建问题 physically plausible

🔬 支柱五：交互与反应 (Interaction & Reaction) (2 篇)

#题目一句话要点标签🔗⭐
96 LongEgoRefer: A Benchmark for Long-Form Egocentric Video Referring Expression Comprehension 提出LongEgoRefer以解决长视频自我中心引用表达理解问题 human-object interaction egocentric Ego4D
97 InterCMDM: Block-Causal Diffusion for Autoregressive Human Interaction Generation 提出InterCMDM以解决人际互动生成中的因果性问题 two-person interaction

🔬 支柱八：物理动画 (Physics-based Animation) (1 篇)

#题目一句话要点标签🔗⭐
98 From One-to-One to Many-to-Many: Dynamic Cross-Layer Injection for Deep Vision-Language Fusion 提出动态跨层注入框架以解决视觉语言模型的特征瓶颈问题 AMP large language model multimodal

🔬 支柱七：动作重定向 (Motion Retargeting) (1 篇)

#题目一句话要点标签🔗⭐
99 SAMoR: Motion Modelling for Articulated Objects of Any Skeleton and Topology 提出SAMoR以解决任意骨架拓扑的运动建模问题 motion representation

⬅️ 返回 cs.CV 首页 · 🏠 返回主页