cs.CV（2026-10-29）

📊 共 54 篇论文 | 🔗 15 篇有代码

🎯 兴趣领域导航

支柱二：RL算法与架构 (RL & Architecture) (23 🔗5) 支柱九：具身大模型 (Embodied Foundation Models) (14 🔗4) 支柱三：空间感知与语义 (Perception & Semantics) (8 🔗5) 支柱一：机器人控制 (Robot Control) (6) 支柱六：视频提取与匹配 (Video Extraction) (3 🔗1)

🔬 支柱二：RL算法与架构 (RL & Architecture) (23 篇)

#题目一句话要点标签🔗⭐
1 LucidLyra: Feed-Forward Generative 3D Scene Reconstruction and Joint Understanding via Multi-Task Distillation 提出LucidLyra以解决3D场景重建与理解问题 distillation 3DGS 3D reconstruction
2 WorldFlow: Joint Video and 3D Motion Prediction via Large-Scale Pretraining 提出WorldFlow以解决未来物理状态预测问题 world model world models motion prediction ✅
3 Benchmarking color judgment in multimodal LLMs with CIE-computable ground truth 提出基于CIE色彩标准的多模态LLM色彩判断基准 MAE large language model multimodal
4 MRC-MMAgent: Reinforcement Learning for Multi-Role Collaboration in Multimodal Tool-Using Agents 提出MRC-MMAgent以解决多角色协作的强化学习问题 reinforcement learning multimodal
5 WorldProbe: When Video World Models Fall Apart under Long-Horizon Spatial Exploration 提出WorldProbe以解决长时间空间探索中的视频世界模型问题 world model world models multimodal ✅
6 Fight World Model: Customizable Combat through Visual Interaction Proxies 提出Fight World Model以解决游戏战斗一致性问题 world model world models
7 Video Generation Meets Agent Harness: Agentic Video Generation, Agentic World Modeling and Recursive Self-Improvement 提出代理视频生成框架以解决视频生成的可控性和反馈问题 world model world models
8 SelfThinker: Self-Evolving Reasoning via Reliability-Verified Intrinsic Visual Thoughts 提出SelfThinker以解决视觉理解中的人类注释依赖问题 flow matching distillation multimodal ✅
9 GROME: GROunding Factorized MEmory into Video World Models 提出GROME以解决视频世界模型中的长时间一致性问题 world model world models
10 Soil Moisture Estimation in Precision Agriculture: A Knowledge Distillation Approach 提出知识蒸馏方法以提升精准农业中的土壤湿度估计 distillation multimodal
11 LoRe-MIM: Lost-or-Replaced Masked Image Modeling for Self-Supervised Representation Learning in Whole-Body PET/CT 提出LoRe-MIM以解决全身PET/CT图像自监督学习问题 representation learning multimodal
12 Segment Anything Model Based Context Guided Mamba Unet for Medical Image Segmentation 提出SAMBA-UNet以解决医学图像分割中的上下文理解与细节保留问题 Mamba SSM state space model
13 sDMD: Sinkhorn Distribution Matching Distillation 提出sDMD以解决DMD在样本引导中的局限性 distillation
14 YOLOF-MLD: A Multi-Level Distillation Framework for Efficient Object Detection 提出YOLOF-MLD框架以提升边缘设备的目标检测效率 distillation ✅
15 DeCoRL: Decomposed Consistency Reinforcement Learning for Multi-Image Composition 提出DeCoRL以解决多图像合成一致性问题 reinforcement learning
16 L3G-World: Real-Time Infinite World Generation with Latent 3D Gaussian Memory 提出L3G-World以解决实时无限世界生成中的空间一致性问题 world model world models
17 Wizard: Manipulate Any World in Real Time via Agentic Control 提出Wizard以解决实时虚拟世界控制不足的问题 world model world models
18 FutureWorld: A Plug-and-Play Causal Next-State Predictor for Interactive Video World Exploration 提出Causal Next-Chunk Predictor以解决交互视频生成中的错误累积问题 world model world models ✅
19 Robust LIDAR Sensor-to-Sensor Domain Generalization via Accumulated-Point Scene Completion 提出基于占用预测的LIDAR传感器领域泛化方法以解决数据迁移问题 world model world models
20 WebOverlay: High-Throughput Vision-Only Distillation for Extreme Aspect-Ratio Web Overlay Detection 提出WebOverlay以解决极端纵横比网页覆盖物检测问题 distillation large language model multimodal
21 Dreams Are Cheaper Than Flights: Curated World-Model Data Improves Aerial Detection 利用合成数据提升无人机野生动物检测性能 world model world models
22 Let LiDAR Decide Where to Look: Geometry-Measured Token Selection for Efficient Sparse Latent World Models 提出几何感知令牌选择以提升稀疏潜在世界模型的效率 world model world models
23 GROME: GROunding Factorized MEmory into Video World Models 提出GROME以解决视频世界模型中的长时间一致性问题 world model world models

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

#题目一句话要点标签🔗⭐
24 Diagnosing the Reference Gap: What Visual Scaffolds Can and Cannot Fix in Multimodal Reasoning 提出ReGap-Bench以解决多模态推理中的参考差距问题 large language model multimodal
25 RepConTUR: A Report-Conditioned Volumetric Multimodal Model for Abdominal Tumor Segmentation 提出RepConTUR以解决腹部肿瘤自动分割中的报告依赖问题 multimodal ✅
26 DiFuse: Fusing Native Diffusion Representations for Spatial Reasoning in Unified Multimodal Models 提出DiFuse以增强统一多模态模型的空间推理能力 multimodal
27 MemoryFusion: Cross-Temporal Memory Learning for Multimodal Video Fusion 提出MemoryFusion以解决多模态视频融合中的长短期信息不足问题 multimodal ✅
28 PhysCount: Do MLLMs Count Visual Appearances or Physical Entities? 提出PhysCount以解决视觉外观与物理实体计数问题 large language model multimodal
29 Radiology Guideline-derived Decision Graph for Efficient Clinical Decision Support 提出GDGraph以解决医学影像临床决策支持问题 large language model chain-of-thought
30 VideoMemBench: How Well Do MLLMs Remember in Video Conversations? 提出VideoMemBench以评估多模态大语言模型在视频对话中的记忆能力 large language model multimodal
31 Mind The Gap (MTG-AD): Dataset and Fine-grained Evaluation for End-to-End Inline Audio Descriptions 提出MTG-AD以解决盲人用户的音频描述生成问题 large language model multimodal ✅
32 Spectral Invariant Fusion and Refinement for Domain Generalized Semantic Segmentation 提出SIFR以解决领域泛化语义分割问题 foundation model
33 Evaluating Cultural Knowledge Editing in Vision-Language Models 提出CulturalBind以解决视觉语言模型中的文化知识编辑问题 multimodal
34 Calibrating Attention-Based Token Pruning for Efficient Multi-Image Synthesis in UMMs 提出CATP方法以解决UMMs中多图像合成的效率问题 multimodal
35 CAD-WAM: Anticipating Geometry for Text-to-CAD Generation 提出CAD-WAM以解决文本到CAD生成中的几何预测问题 large language model ✅
36 Sparse Autoencoders for Expert-Assisted Interpretation of Histopathology Foundation Model Embeddings 提出稀疏自编码器以辅助解读病理学基础模型嵌入 foundation model
37 FieldGuideRAG: Image-Grounded GraphRAG for Fine-Grained Recognition with Vision-Language Models 提出FieldGuideRAG以解决细粒度物种识别问题 multimodal

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

#题目一句话要点标签🔗⭐
38 MLLM-GS: Multimodal Large Language Model Prior-Guided Fast Novel View Synthesis in Extreme Environments 提出MLLM-GS以解决极端环境下的新视图合成问题 3DGS 3D reconstruction scene reconstruction ✅
39 ACD-GS: Asymmetric Curvature-aware Densification for 3D Gaussian Splatting 提出ACD-GS以解决3D Gaussian Splatting训练中的冗余问题 3D gaussian splatting 3DGS gaussian splatting ✅
40 COGS: A Diagnostic Benchmark for Attribute-Aware Object-Level 3D Gaussian Splatting Perception 提出COGS数据集以解决3D感知中高斯属性利用不均的问题 3D gaussian splatting 3DGS gaussian splatting ✅
41 PhysInsert: Physically Plausible Object Insertion with Geometric Priors and Bidirectional Object-Scene Interactions 提出PhysInsert以解决物体插入中的物理一致性问题 VGGT physically plausible
42 PriorMaskMap: Robust Online Vectorized Map Construction with Biased Priors 提出PriorMaskMap以解决在线向量化地图构建中的先验偏差问题 scene understanding ✅
43 EgoHSI3R: Online 4D Hand-Scene Interaction Reconstruction from Egocentric Video 提出EgoHSI3R以解决手-场景交互重建问题 scene reconstruction MANO egocentric
44 Prior-Guided Single-Exposure HDR Novel View Synthesis via Base-Gain Representation 提出PriorGS以解决单曝光HDR新视图合成问题 3D gaussian splatting gaussian splatting splatting
45 Inter-Frame Predictive Contour Coding for Semantic Map Compression 提出跨帧预测轮廓编码以解决动态语义地图压缩问题 semantic map ✅

🔬 支柱一：机器人控制 (Robot Control) (6 篇)

#题目一句话要点标签🔗⭐
46 HiFOR: Hierarchical Foresight with Action-Guided World Grounding for Vision-Language-Action Model 提出HiFOR框架以解决VLA模型对未来预测的不足问题 manipulation world model world models
47 Seeing and Imagining as One: Unified Video Generation and 4D Perception for Embodied World Modeling 提出GenSight以解决未来视频生成与4D感知统一问题 manipulation world model world models
48 RFW-A: Agentic Procedural Generation of 400K Articulated Assets over 500 Categories 提出RFW-A以高效生成40万类人形资产解决构建难题 manipulation
49 Seeing and Imagining as One: Unified Video Generation and 4D Perception for Embodied World Modeling 提出GenSight以解决未来视频生成与4D感知统一问题 manipulation world model world models
50 DyCraft: An Agentic System for Long-Horizon 4D Scene Generation 提出DyCraft以解决长时间动画生成中的物理不一致问题 humanoid humanoid robot spatiotemporal
51 RADI3N: GeometRic ImAge EDiting via ParametrIc 3D CoNtrol 提出RADI3N以解决图像编辑中的3D控制问题 manipulation

🔬 支柱六：视频提取与匹配 (Video Extraction) (3 篇)

#题目一句话要点标签🔗⭐
52 Probing Spatial State Representations in Visual Foundation Models 提出空间状态表示方法以提升视觉基础模型的空间感知能力 egocentric foundation model
53 Hierarchical Step Decomposition for Training-free Online Video Step Grounding 提出分层步骤分解方法以解决在线视频步骤定位问题 Ego4D multimodal
54 TactoHand: Learning Geometry-Based Dense Hand-Surface Interaction Fields from Video 提出TactoHand以解决手部表面交互场景理解问题 hand-object reconstruction ✅

⬅️ 返回 cs.CV 首页 · 🏠 返回主页