cs.CV（2024-10-17）

📊 共 28 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) (11 🔗2) 支柱三：空间感知与语义 (Perception & Semantics) (7 🔗3) 支柱二：RL算法与架构 (RL & Architecture) (3) 支柱一：机器人控制 (Robot Control) (2 🔗1) 支柱八：物理动画 (Physics-based Animation) (2) 支柱四：生成式动作 (Generative Motion) (2 🔗1) 支柱五：交互与反应 (Interaction & Reaction) (1 🔗1)

🔬 支柱九：具身大模型 (Embodied Foundation Models) (11 篇)

#题目一句话要点标签🔗⭐
1 $γ-$MoD: Exploring Mixture-of-Depth Adaptation for Multimodal Large Language Models 提出$γ$-MoD，通过深度混合自适应提升多模态大语言模型的效率。 large language model multimodal
2 RAP: Retrieval-Augmented Personalization for Multimodal Large Language Models RAP：检索增强的个性化多模态大语言模型框架 large language model multimodal ✅
3 Improving Multi-modal Large Language Model through Boosting Vision Capabilities Arcana：通过增强视觉能力提升多模态大语言模型性能 large language model multimodal
4 Fundus to Fluorescein Angiography Video Generation as a Retinal Generative Foundation Model 提出Fundus2Video，用于从眼底彩照生成动态FFA视频，并作为视网膜生成式基础模型。 foundation model multimodal
5 Janus: Decoupling Visual Encoding for Unified Multimodal Understanding and Generation Janus：解耦视觉编码，实现统一的多模态理解与生成 multimodal
6 Movie Gen: A Cast of Media Foundation Models Movie Gen：一套高质量媒体基础模型，实现1080p高清视频生成与编辑 foundation model
7 Temporal-Enhanced Multimodal Transformer for Referring Multi-Object Tracking and Segmentation 提出TenRMOT，利用时序增强Transformer解决Referring多目标跟踪与分割问题 multimodal
8 Harnessing Webpage UIs for Text-Rich Visual Understanding 利用网页UI提升富文本视觉理解能力，解决多模态大模型在结构化环境中的交互问题。 large language model multimodal
9 Exploring the Design Space of Visual Context Representation in Video MLLMs 提出视觉上下文表示设计方案以提升视频多模态大语言模型性能 large language model multimodal ✅
10 Performance of Gaussian Mixture Model Classifiers on Embedded Feature Spaces 提出一种参数量更少的GMM分类器，并评估其在CLIP和ImageBind嵌入特征空间上的性能。 multimodal
11 Trust but Verify: Programmatic VLM Evaluation in the Wild 提出程序化VLM评估方法以解决视觉语言模型的响应验证问题 large language model

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

#题目一句话要点标签🔗⭐
12 GlossyGS: Inverse Rendering of Glossy Objects with 3D Gaussian Splatting GlossyGS：利用3D高斯溅射和材质先验进行光泽物体逆渲染 3D gaussian splatting gaussian splatting splatting
13 DepthSplat: Connecting Gaussian Splatting and Depth DepthSplat：连接高斯溅射与深度估计，实现高质量三维重建与深度预测。 depth estimation monocular depth 3D gaussian splatting
14 MEGA: Memory-Efficient 4D Gaussian Splatting for Dynamic Scenes MEGA：面向动态场景的内存高效4D高斯溅射方法 gaussian splatting splatting ✅
15 VLM-Grounder: A VLM Agent for Zero-Shot 3D Visual Grounding VLM-Grounder：一种基于视觉语言模型的零样本3D视觉定位方法 scene understanding visual grounding ✅
16 DN-4DGS: Denoised Deformable Network with Temporal-Spatial Aggregation for Dynamic Scene Rendering 提出DN-4DGS，通过去噪和时空聚合实现动态场景实时高质量渲染 3D gaussian splatting 3DGS gaussian splatting
17 ARKit LabelMaker: A New Scale for Indoor 3D Scene Understanding ARKit LabelMaker：构建大规模室内3D场景理解数据集，提升语义分割性能 scene understanding ✅
18 Self-Supervised Scene Flow Estimation with Point-Voxel Fusion and Surface Representation 提出点-体素融合与表面表示的自监督场景流估计方法，提升三维运动场预测精度。 scene flow

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

#题目一句话要点标签🔗⭐
19 DriveDreamer4D: World Models Are Effective Data Machines for 4D Driving Scene Representation DriveDreamer4D：利用世界模型作为数据机器，提升4D驾驶场景表示 world model dreamer 3DGS
20 Stochastic Flow Matching for Resolving Small-Scale Physics 提出随机流匹配（SFM）框架，用于解决物理科学中小尺度细节的超分辨率重建问题。 flow matching
21 Enhancing Dataset Distillation via Label Inconsistency Elimination and Learning Pattern Refinement M-DATM：通过消除标签不一致性和优化学习模式提升数据集蒸馏效果 distillation

🔬 支柱一：机器人控制 (Robot Control) (2 篇)

#题目一句话要点标签🔗⭐
22 Object Pose Estimation Using Implicit Representation For Transparent Objects 提出基于NeRF隐式表达的透明物体位姿估计方法，超越现有技术水平。 manipulation NeRF neural radiance field
23 PUMA: Empowering Unified MLLM with Multi-granular Visual Generation PUMA：提出一种统一的多模态大语言模型，赋能多粒度视觉生成任务。 manipulation large language model foundation model ✅

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

#题目一句话要点标签🔗⭐
24 Spatiotemporal Object Detection for Improved Aerial Vehicle Detection in Traffic Monitoring 提出时空车辆检测模型，提升无人机交通监控中车辆检测精度 spatiotemporal
25 Training Compute-Optimal Vision Transformers for Brain Encoding 针对大脑编码，研究计算量最优的视觉Transformer训练策略 spatiotemporal

🔬 支柱四：生成式动作 (Generative Motion) (2 篇)

#题目一句话要点标签🔗⭐
26 MotionBank: A Large-scale Video Motion Benchmark with Disentangled Rule-based Annotations MotionBank：构建大规模视频动作基准，用于解耦的规则驱动型动作描述生成。 motion generation foundation model ✅
27 L3DG: Latent 3D Gaussian Diffusion L3DG：提出基于潜空间3D高斯扩散的生成式3D建模方法，可扩展到房间级场景。 VQ-VAE

🔬 支柱五：交互与反应 (Interaction & Reaction) (1 篇)

#题目一句话要点标签🔗⭐
28 GraspDiffusion: Synthesizing Realistic Whole-body Hand-Object Interaction GraspDiffusion：合成逼真全身人-物交互场景 human-object interaction ✅

⬅️ 返回 cs.CV 首页 · 🏠 返回主页