cs.CV（2024-04-30）

📊 共 25 篇论文 | 🔗 8 篇有代码

🎯 兴趣领域导航

支柱九：具身大模型 (Embodied Foundation Models) (8 🔗2) 支柱三：空间感知与语义 (Perception & Semantics) (7 🔗2) 支柱四：生成式动作 (Generative Motion) (3 🔗1) 支柱二：RL算法与架构 (RL & Architecture) (3 🔗2) 支柱八：物理动画 (Physics-based Animation) (2) 支柱一：机器人控制 (Robot Control) (1 🔗1) 支柱六：视频提取与匹配 (Video Extraction) (1)

🔬 支柱九：具身大模型 (Embodied Foundation Models) (8 篇)

#题目一句话要点标签🔗⭐
1 Naturally Supervised 3D Visual Grounding with Language-Regularized Concept Learners 提出语言正则化概念学习器以解决自然监督3D视觉定位问题 large language model visual grounding
2 Large Language Model Informed Patent Image Retrieval 提出语言信息驱动的多模态方法以解决专利图像检索问题 large language model multimodal
3 Training a high-performance retinal foundation model with half-the-data and 400 times less compute 提出RETFound-Green以解决数据稀缺与计算资源消耗问题 foundation model
4 Seeing Through the Clouds: Cloud Gap Imputation with Prithvi Foundation Model 提出基于ViT模型的云像素填补方法以解决卫星图像缺失问题 foundation model
5 Revisiting the Adversarial Robustness of Vision Language Models: a Multimodal Perspective 提出多模态对抗训练方法以提升视觉语言模型的鲁棒性 multimodal ✅
6 Transcrib3D: 3D Referring Expression Resolution through Large Language Models 提出Transcrib3D以解决3D指称表达解析问题 large language model ✅
7 Visual Fact Checker: Enabling High-Fidelity Detailed Caption Generation 提出VisualFactChecker以解决自动图像字幕生成的细节不足问题 large language model instruction following
8 VimTS: A Unified Video and Image Text Spotter for Enhancing the Cross-domain Generalization 提出VimTS以解决跨域文本检测问题 multimodal

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

#题目一句话要点标签🔗⭐
9 MoST: Multi-modality Scene Tokenization for Motion Prediction 提出多模态场景标记方法以解决运动预测中的感知误差问题 open-vocabulary open vocabulary motion prediction
10 RTG-SLAM: Real-time 3D Reconstruction at Scale using Gaussian Splatting 提出RTG-SLAM以解决大规模环境下实时3D重建问题 3D reconstruction gaussian splatting splatting
11 GS-LRM: Large Reconstruction Model for 3D Gaussian Splatting 提出GS-LRM以解决3D重建中的高效性与复杂性问题 3D gaussian splatting gaussian splatting splatting ✅
12 NeRF-Insert: 3D Local Editing with Multimodal Control Signals 提出NeRF-Insert以解决3D场景局部编辑问题 NeRF multimodal
13 One-Stage Open-Vocabulary Temporal Action Detection Leveraging Temporal Multi-scale and Action Label Features 提出单阶段开放词汇时间动作检测方法以解决现有方法的局限性 open-vocabulary open vocabulary
14 Lightplane: Highly-Scalable Components for Neural 3D Fields 提出Lightplane以解决2D-3D映射中的内存瓶颈问题 3D reconstruction ✅
15 Masked Spatial Propagation Network for Sparsity-Adaptive Depth Refinement 提出稀疏自适应深度精细化框架以解决深度补全问题 monocular depth

🔬 支柱四：生成式动作 (Generative Motion) (3 篇)

#题目一句话要点标签🔗⭐
16 MotionLCM: Real-time Controllable Motion Generation via Latent Consistency Model 提出MotionLCM以解决实时可控运动生成效率问题 motion generation motion latent MotionLCM
17 Fake it to make it: Using synthetic data to remedy the data shortage in joint multimodal speech-and-gesture synthesis 提出合成数据方法以解决多模态语音与手势合成的数据短缺问题 motion synthesis multimodal ✅
18 A Minimal Set of Parameters Based Depth-Dependent Distortion Model and Its Calibration Method for Stereo Vision Systems 提出基于最小参数集的深度依赖畸变模型以提升立体视觉系统精度 MDM

🔬 支柱二：RL算法与架构 (RL & Architecture) (3 篇)

#题目一句话要点标签🔗⭐
19 MicroDreamer: Efficient 3D Generation in $\sim$20 Seconds by Score-based Iterative Reconstruction 提出MicroDreamer以解决3D生成效率低下问题 dreamer distillation 3D gaussian splatting ✅
20 CLIP-Mamba: CLIP Pretrained Mamba Models with OOD and Hessian Evaluation 提出CLIP-Mamba模型以提升零-shot分类与OOD泛化能力 Mamba state space model ✅
21 Causal Perception Inspired Representation Learning for Trustworthy Image Quality Assessment 提出因果感知启发的表示学习以解决图像质量评估问题 representation learning

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

#题目一句话要点标签🔗⭐
22 EvGNN: An Event-driven Graph Neural Network Accelerator for Edge Vision 提出EvGNN以解决边缘视觉中的事件驱动图神经网络加速问题 spatiotemporal
23 Cross-Block Fine-Grained Semantic Cascade for Skeleton-Based Sports Action Recognition 提出跨块细粒度语义级联模块以解决骨架基础运动动作识别问题 spatiotemporal

🔬 支柱一：机器人控制 (Robot Control) (1 篇)

#题目一句话要点标签🔗⭐
24 PACER+: On-Demand Pedestrian Animation Controller in Driving Scenarios 提出PACER+以解决行驶场景中行人动画多样性与可控性问题 motion tracking human motion ✅

🔬 支柱六：视频提取与匹配 (Video Extraction) (1 篇)

#题目一句话要点标签🔗⭐
25 Ultra Inertial Poser: Scalable Motion Capture and Tracking from Sparse Inertial Sensors and Ultra-Wideband Ranging 提出Ultra Inertial Poser以解决稀疏传感器运动捕捉精度问题 sparse sensors human motion

⬅️ 返回 cs.CV 首页 · 🏠 返回主页