cs.CV（2026-09-22）

📊 共 36 篇论文 | 🔗 4 篇有代码

🎯 兴趣领域导航

支柱二：RL算法与架构 (RL & Architecture) (15 🔗3) 支柱九：具身大模型 (Embodied Foundation Models) (10) 支柱三：空间感知与语义 (Perception & Semantics) (7 🔗1) 支柱一：机器人控制 (Robot Control) (4)

🔬 支柱二：RL算法与架构 (RL & Architecture) (15 篇)

#题目一句话要点标签🔗⭐
1 Latent Dataset Distillation for Human Motion Prediction 提出潜在数据蒸馏框架以解决人类运动预测中的不稳定性问题 distillation VQ-VAE human motion
2 BAS-OPD: Budget-Aware Selective On-Policy Self-Distillation for Fine-Grained Multimodal Perception 提出BAS-OPD以解决多模态视觉感知中的监督成本问题 distillation large language model multimodal
3 Agentic Building-Aware Satellite Gaussian Splatting for Auditable Urban DSM Reconstruction 提出建筑感知卫星高斯点云重建方法以提升城市DSM重建精度 MAE 3D reconstruction gaussian splatting
4 MMAP: Multimodal Missing-Aware Pretraining for Longitudinal Alzheimer's Prediction 提出MMAP以解决多模态医疗数据缺失问题 contrastive learning foundation model multimodal
5 ForeDrive: Foresight-Guided End-to-End Autonomous Driving with a Planning-Relevant Latent World Model 提出ForeDrive以解决自主驾驶中的规划相关表示问题 reinforcement learning imitation learning world model
6 GAD-MambaUNet: Direction-Group Mamba with Gradient-Adaptive DINOv3 Distillation for Lightweight Medical Image Segmentation 提出GAD-MambaUNet以解决轻量级医学图像分割问题 Mamba distillation foundation model
7 GeoComposer: Geometry-Grounded Photographic Composition Instruction 提出GeoComposer以解决摄影构图中几何一致性问题 reinforcement learning representation learning geometric consistency
8 Point Diffusion Mamba: Unified Diffusion-State-Space Modeling for Single-View 3D Reconstruction under Data Scarcity 提出Point Diffusion Mamba以解决数据稀缺下的单视图3D重建问题 Mamba 3D reconstruction ✅
9 Code Plans, Diffusion Renders: Open-Ended Generative World Modeling 提出CoDeR以解决开放式生成世界建模问题 world model world models ✅
10 QuantWM: Temporally Consistent 2-Bit KV Cache Quantization for World Models and Video Generation 提出QuantWM以解决KV缓存量化中的视觉退化问题 world model world models
11 GameDirector: Decoupling Gameplay Logic from Rendering for Player-Configurable Game World Models 提出GameDirector以解决游戏逻辑与渲染耦合问题 world model world models
12 DIFTA-3D: Depth-Consistent Instance-Level Feature Transfer and Adaptation of DINOv3 for 3D Detection 提出DIFTA-3D以解决RGB-D 3D实例检测中的特征依赖问题 distillation metric depth foundation model
13 KwaiMind Technical Report 提出KwaiMind以解决商业图像编辑中的产品一致性问题 reinforcement learning distillation multimodal
14 GTR: Gated Token Recurrence for Efficient Dense Prediction 提出GTR以解决高分辨率密集预测的效率问题 linear attention depth estimation monocular depth ✅
15 Video-HopChain: Multi-Hop Questions and Confidence-Gated Exploration for Video Reasoning Models 提出Video-HopChain以解决视频推理中的多跳问题 reinforcement learning chain-of-thought

🔬 支柱九：具身大模型 (Embodied Foundation Models) (10 篇)

#题目一句话要点标签🔗⭐
16 From Token Importance to Conditional Removability: Rethinking Visual Token Pruning in Multimodal Large Language Models 提出CoRePrune以解决视觉标记修剪中的条件可移除性问题 large language model multimodal
17 Vision Foundation Models with Synthetic-Only Training for Monocular Spacecraft Pose Estimation 提出基于合成数据的视觉基础模型以解决单目航天器姿态估计问题 foundation model
18 MatchFusion: Explicit-Implicit Instance Matching for Spatio-Temporal Multimodal Autonomous Driving 提出MatchFusion以解决多模态自动驾驶中的实例匹配问题 multimodal
19 ROAM-ASD: Robust Open-World Active Speaker Detection with Flexible Multimodal Fusion 提出ROAM-ASD以解决主动说话者检测中的鲁棒性问题 multimodal
20 Foundation model embeddings capture pre-diagnostic changes on screening mammograms 提出基础模型嵌入以捕捉筛查乳腺X光片上的前诊断变化 foundation model
21 Virtual Encoders in Multimodal Transformers 提出虚拟编码器以优化多模态变换器的感知处理 multimodal
22 PartLLM: A Unified Multimodal Foundation for 3D Part Segmentation 提出PartLLM以统一解决3D部件分割问题 multimodal
23 When Point Clouds Outperform Pixels: Rethinking Zero-Shot Multimodal Anomaly Detection 提出WOOPS框架以解决零-shot多模态异常检测中的模态可靠性问题 multimodal
24 RootQuantV2: Adapting a Vision Foundation Model for Root-Trait Regression from Minirhizotron Imagery 提出RootQuantV2以解决根系性状自动估计问题 foundation model
25 Calibrating Retrieval Geometry: Reliability-Guided Training-Free Aggregation for Visual Place Recognition 提出TFA方法以解决视觉场所识别中的聚合问题 foundation model

🔬 支柱三：空间感知与语义 (Perception & Semantics) (7 篇)

#题目一句话要点标签🔗⭐
26 Ultra-fast Neural Inference for Stochastic Gaussian Splatting Denoising 提出时序神经去噪方法以解决高斯点云渲染中的噪声问题 gaussian splatting splatting
27 HARMONY: Hierarchical Agentic Reasoning for MONocular Image-to-Scene Synthesis 提出HARMONY以解决单目图像到场景合成的挑战 scene reconstruction spatial relationship foundation model
28 Evidence-gated multimodal parsing and vectorization of architectural floor plans 提出SALI-FP以解决建筑平面图数字化难题 semantic map multimodal
29 NaCR: Visual Localization via NeRF-aided Camera Ray Regression 提出NeRF辅助的相机光线回归以提升视觉定位精度 NeRF neural radiance field
30 Robust, Estimator-Agnostic Dynamic 3DGS Compression 提出一种稳健的动态3D高斯点云压缩方法以解决时间冗余问题 3DGS ✅
31 LiFR v2: Completion-Augmented Event Propagation for High-Rate Dense Prediction 提出LiFR v2以解决动态环境中高频率密集预测问题 depth estimation monocular depth
32 GRIP: Gaussian Rendering as a Cross-Modal Bridge for Image-to-Point Cloud Registration 提出GRIP框架以解决图像与点云配准问题 splatting

🔬 支柱一：机器人控制 (Robot Control) (4 篇)

#题目一句话要点标签🔗⭐
33 Fysiverse-3D-Vision Technical Report: Generating Executable 3D Worlds from Images through Unified Spatial Reasoning 提出Fysiverse-3D-Vision以解决从单幅图像生成可执行3D场景的问题 manipulation scene reconstruction spatial relationship
34 Metric-Bench: Exploring In-context Spatial Metric Reasoning in VLMs for Indoor Scenes 提出Metric-Bench以解决视觉语言模型中的空间度量推理问题 manipulation embodied AI multimodal
35 Semantically-Guided Domain Randomization for Industrial Object Detection in Low-Image-Budget Regimes 提出语义引导的领域随机化以解决低图像预算下的工业物体检测问题 domain randomization
36 TRACE: Trajectory Representation and Consistency Estimation for AI-Generated Video Detection 提出TRACE以解决AI生成视频检测中的一致性问题 trajectory optimization flow matching representation learning

⬅️ 返回 cs.CV 首页 · 🏠 返回主页