cs.CV（2026-10-14）

📊 共 51 篇论文 | 🔗 9 篇有代码

🎯 兴趣领域导航

支柱二：RL算法与架构 (RL & Architecture) (16 🔗4) 支柱三：空间感知与语义 (Perception & Semantics) (15 🔗5) 支柱九：具身大模型 (Embodied Foundation Models) (13) 支柱六：视频提取与匹配 (Video Extraction) (2) 支柱八：物理动画 (Physics-based Animation) (2) 支柱四：生成式动作 (Generative Motion) (1) 支柱七：动作重定向 (Motion Retargeting) (1) 支柱一：机器人控制 (Robot Control) (1)

🔬 支柱二：RL算法与架构 (RL & Architecture) (16 篇)

#题目一句话要点标签🔗⭐
1 OmniPoser: Flexible Human Motion Recovery in the Wild with Masked Flow Matching 提出OmniPoser以解决稀疏传感器下的人体运动恢复问题 flow matching SMPL human motion ✅
2 MMKG-R1: Constructing Multimodal Knowledge Graphs via Reinforcement Learning 提出MMKG-R1以解决多模态知识图谱构建质量不足的问题 reinforcement learning large language model multimodal
3 SonoWorld: Towards Interactive Pose-Controllable Ultrasound World Models 提出SonoWorld以解决超声图像控制环境不足问题 policy learning world model world models
4 Reliability-Aware Multimodal Sentiment Prediction Under Local Modality Missingness 提出可靠性意识的多模态情感预测以应对局部模态缺失问题 predictive model multimodal
5 Multimodal Medical Image Parsing with Task-Specific Low-Rank Adapters 提出任务特定低秩适配器以解决多模态医学图像解析问题 MAE multimodal ✅
6 PAWA: Panoramic Action-conditioned World Modeling for Multiple Agents 提出PAWA以解决多玩家环境建模问题 world model world models
7 StateCraft: Augmenting Video World Models with Abstract-State Reasoning 提出StateCraft以解决视频世界模型的状态一致性问题 world model world models ✅
8 Glance Forcing: Making Bidirectional Diffusion Autoregressive at a Glance 提出Glance Forcing以解决双向扩散模型的高成本问题 world model world models distillation
9 SAM3RL: Memory Control via Reinforcement Learning for Visual Object Tracking 提出SAM3RL通过强化学习优化视觉目标跟踪的内存控制 reinforcement learning foundation model
10 FPV-CRASH: A Dataset for Offline Crash Moment Localization in Drone First-Person-View Videos 提出FPV-CRASH数据集以解决无人机FPV视频中的碰撞时刻定位问题 MAE first-person view
11 Learning Visual Speech Representations via Cross-Modal Distillation and Joint Face-Lip Modeling 提出FLARE框架以解决视觉语音识别中的模糊性问题 representation learning distillation ✅
12 RepForcing: Representation-Space Initialization for Autoregressive Video Distillation 提出RepForcing以解决短步长自回归视频蒸馏生成质量问题 distillation
13 Player-1: Learning to Play from Synthetic Experience 提出Player-1框架以解决游戏控制学习的挑战 reinforcement learning PPO
14 OneLA: Learning Many Worlds from Latent Actions 提出OneLA框架以解决跨域动作学习问题 world model world models
15 BluePixel: Toward Native 4K Agentic Image Editing 提出BluePixel以解决4K图像编辑训练难题 reinforcement learning distillation
16 SceneAE: Scene Generation and Reconstruction in Queryable 3D-Anchored Latent Space 提出SceneAE以解决3D场景生成与重建问题 world model world models

🔬 支柱三：空间感知与语义 (Perception & Semantics) (15 篇)

#题目一句话要点标签🔗⭐
17 VisibilitySplat: Ray-Traced Occlusion for Relighting with 3D Gaussian Splatting 提出VisibilitySplat以解决逆向渲染中的可见性建模问题 3D gaussian splatting 3DGS gaussian splatting
18 Feed-Forward Multi-View Feature Coding for 3D Gaussian Splatting 提出多视图特征编码以解决3D高斯点云重建问题 3D gaussian splatting 3DGS gaussian splatting
19 RDCompGS: Rate-Distortion Optimization for Compression of 3D Gaussian Splatting with Image Codecs 提出RDCompGS以解决3D高斯点云压缩中的率失真优化问题 3D gaussian splatting 3DGS gaussian splatting
20 Why 3D Reconstruction Breaks Down on Under-Ice AUV Video, and How to Predict It 提出检测系统以解决水下AUV视频三维重建失败问题 metric depth 3D reconstruction splatting
21 Where Should the Gaussians Go? Reusable Label-Aware Allocation for Budget-Constrained Gaussian Splatting of Cinematic CT 提出可重用的标签感知分配方法以优化预算约束下的高斯点云 3D gaussian splatting gaussian splatting splatting
22 HyperLoop: Hyperspectral-Assisted Loop Closure Detection in 3D-SLAM for Planetary Exploration 提出HyperLoop以解决行星探测中的回环闭合问题 3D gaussian splatting 3DGS gaussian splatting
23 Holi-Spatial 360: Generative Reconstruction from Sparse Panoramas for Spatial Rendering and Intelligence 提出Holi-Spatial 360以解决稀疏全景重建问题 3D gaussian splatting 3DGS gaussian splatting
24 Remember, Revisit, Revise: Scaffold Memory for Long-Horizon 3D Geometry 提出SMORE框架以解决长序列3D几何重建问题 depth estimation scene reconstruction foundation model
25 MicroSDF: Microfacet-Driven Hybrid Neural SDFs for Mixed-Reflectance Surface Reconstruction 提出MicroSDF以解决混合反射表面重建问题 3D reconstruction NeRF physically plausible
26 NavAble: A Large-Scale Dataset and Synthetic Data Generation Pipeline for Blind Navigation 提出NavAble以解决盲人导航数据匮乏问题 sam 3D SAM 3D egocentric ✅
27 InfiniSplat v2: Analyzing and Improving Feed-Forward 3DGS from Any Views 提出InfiniSplat v2以解决高质量3D场景重建问题 3DGS ✅
28 MagicGarment: Benchmarking Single-Image Standalone 3D Garment Reconstruction 提出MagicGarment以解决单图像独立3D服装重建问题 3D reconstruction multimodal ✅
29 LiFi: LiDAR Generation from Multi-View Images via Geometric and Semantic Collaborative Guidance 提出LiFi框架以解决多视角图像生成LiDAR点云问题 Depth Anything classifier-free guidance ✅
30 OLiVES: An Outdoor Low-Light Video Benchmark for Enhancement and Segmentation 提出OLiVES数据集以解决户外低光视频处理问题 scene understanding ✅
31 CoDy: Codec Dynamics as Complementary Priors for Unsupervised Temporal Action Localization 提出CoDy框架以解决无监督时序动作定位问题 optical flow

🔬 支柱九：具身大模型 (Embodied Foundation Models) (13 篇)

#题目一句话要点标签🔗⭐
32 SPECTRA-VLA: Base-Fine Hyperspectral Token Transfer at Frozen Vision-Language-Action Interfaces 提出SPECTRA-VLA以解决高光谱成像中的分类精度问题 vision-language-action VLA
33 ActICL: Post-Training for Agentic Multimodal In-Context Learning 提出ActICL以解决多模态上下文学习中的演示检索问题 large language model multimodal
34 Quantifying privacy risks in foundation model representations of brain MRI 量化基础模型在脑MRI表示中的隐私风险 foundation model
35 Instruction Mining From Images: Constructing a Synthetic Dataset for Multimodal Learning ILSP DCVLR Competition Submission 提出一种合成数据集生成管道以支持多模态学习 multimodal
36 ReferFlow: Beyond Final-Target Localization for Complex Referring Expressions 提出ReferFlow以解决复杂指代表达的中间目标定位问题 large language model multimodal
37 Beyond Static Cues: Unified Dynamic Detection of MLLM Backdoors via Dual-Flow Trajectory Analysis 提出EVO-Detect以解决多模态大语言模型后门检测问题 large language model multimodal
38 VisualWorldBench: A Fine-Grained Multi-Task Benchmark for Evaluating Visual World Knowledge in MLLMs 提出VisualWorldBench以解决多模态大语言模型的视觉知识评估问题 large language model multimodal
39 LED-BENCH: A Benchmark for Large-format Engineering Drawing Understanding 提出LED-BENCH以解决大格式工程图理解的挑战 large language model multimodal
40 VLX-Go: Robust Embodied Tracking through Multi-Stage Learning and On-Policy Refinement 提出VLX-Go以解决多目标跟踪中的挑战 multimodal
41 StreamErase: Towards Ultra-Fast Streaming Video Erasing 提出StreamErase以解决视频对象去除的实时性问题 foundation model
42 Mitigating Video Hallucinations via Attention-Guided Representation Perturbation 提出注意力引导的表示扰动以缓解视频幻觉问题 large language model
43 CORE: Contrast-and-Reflect for Self-Supervised Prompt Optimization in Anomaly Synthesis 提出CORE框架以解决工业异常检测样本稀缺问题 multimodal
44 VideoHarness-SRSI: Structured Recursive Self-Improvement for Video Understanding Harness 提出VideoHarness-SRSI以解决视频理解中的自我改进问题 multimodal

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

#题目一句话要点标签🔗⭐
45 DreamHMR: From Video Generation to Human Motion Modeling via Structural Focusing 提出DreamHMR以解决视频中人体运动建模问题 human mesh recovery HMR hand reconstruction
46 Generative World Simulation with DAO and JING 提出生成世界模拟框架以实现持续互动体验 egocentric

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

#题目一句话要点标签🔗⭐
47 GeoThinker-$F$: Endogenous Predictive World Representations for Fast Spatial Intelligence 提出GeoThinker-$F$以解决多模态大语言模型的推理延迟问题 spatiotemporal large language model multimodal
48 Attention-Based Spatio-Temporal Approach for Fire and Smoke Detection 提出基于注意力的时空方法以解决火灾烟雾检测问题 spatiotemporal

🔬 支柱四：生成式动作 (Generative Motion) (1 篇)

#题目一句话要点标签🔗⭐
49 Robust Real-Time Monocular Human Pose Estimation with Insole-Based Sensor Fusion 提出基于传感器融合的单目人类姿态估计以解决视觉遮挡问题 physically plausible human mesh recovery SMPL

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

#题目一句话要点标签🔗⭐
50 Can Learning to Answer Spatial Questions Improve Novel-View Generation? 提出Ask4NV以解决新视图生成中的空间理解问题 geometric consistency multimodal

🔬 支柱一：机器人控制 (Robot Control) (1 篇)

#题目一句话要点标签🔗⭐
51 GeoCloak: An Agentic Framework for Semantics-Preserving Geolocation Privacy via Peripheral Clue Manipulation 提出GeoCloak以解决图像地理位置隐私问题 manipulation

⬅️ 返回 cs.CV 首页 · 🏠 返回主页