cs.CV（2024-12-12）

📊 共 56 篇论文 | 🔗 19 篇有代码

🎯 兴趣领域导航

支柱九：具身大模型 (Embodied Foundation Models) (22 🔗4) 支柱二：RL算法与架构 (RL & Architecture) (14 🔗8) 支柱三：空间感知与语义 (Perception & Semantics) (12 🔗4) 支柱四：生成式动作 (Generative Motion) (2 🔗1) 支柱一：机器人控制 (Robot Control) (2 🔗2) 支柱八：物理动画 (Physics-based Animation) (2) 支柱五：交互与反应 (Interaction & Reaction) (1) 支柱七：动作重定向 (Motion Retargeting) (1)

🔬 支柱九：具身大模型 (Embodied Foundation Models) (22 篇)

#题目一句话要点标签🔗⭐
1 Towards a Multimodal Large Language Model with Pixel-Level Insight for Biomedicine MedPLIB：面向生物医学，具备像素级理解的多模态大语言模型 large language model multimodal ✅
2 EasyRef: Omni-Generalized Group Image Reference for Diffusion Models via Multimodal LLM EasyRef：利用多模态LLM实现扩散模型的多参考图像泛化生成 large language model multimodal instruction following
3 InternLM-XComposer2.5-OmniLive: A Comprehensive Multimodal System for Long-term Streaming Video and Audio Interactions 提出InternLM-XComposer2.5-OmniLive，用于长期流式视频和音频交互的多模态系统 large language model foundation model multimodal
4 Exemplar Masking for Multimodal Incremental Learning 提出Exemplar Masking框架，解决多模态增量学习中的存储和计算瓶颈。 large language model multimodal ✅
5 Agtech Framework for Cranberry-Ripening Analysis Using Vision Foundation Models 提出基于视觉基础模型的蔓越莓成熟度分析框架，用于精准农业 foundation model
6 V2PE: Improving Multimodal Long-Context Capability of Vision-Language Models with Variable Visual Position Encoding 提出V2PE：通过可变视觉位置编码提升视觉-语言模型的多模态长上下文能力 multimodal
7 Multimodal Music Generation with Explicit Bridges and Retrieval Augmentation 提出VMB框架，利用显式桥接和检索增强实现高质量多模态音乐生成 multimodal ✅
8 MaskTerial: A Foundation Model for Automated 2D Material Flake Detection MaskTerial：用于自动二维材料薄片检测的基础模型 foundation model
9 Foundation Models and Adaptive Feature Selection: A Synergistic Approach to Video Question Answering 提出LGQAVE模型，通过自适应特征选择和基础模型增强视频问答性能 foundation model
10 Enriching Multimodal Sentiment Analysis through Textual Emotional Descriptions of Visual-Audio Content 提出DEVA框架，通过文本情感描述增强视听内容的多模态情感分析。 multimodal
11 Pinpoint Counterfactuals: Reducing social bias in foundation models via localized counterfactual generation 提出局部化对抗样本生成方法，降低基础模型中的社会偏见 foundation model
12 ViCaS: A Dataset for Combining Holistic and Pixel-level Video Understanding using Captions with Grounded Segmentation 提出ViCaS数据集以解决视频理解中的高层次与像素级分割问题 large language model multimodal ✅
13 Olympus: A Universal Task Router for Computer Vision Tasks Olympus：一种用于计算机视觉任务的通用任务路由框架 large language model multimodal
14 SynerGen-VL: Towards Synergistic Image Understanding and Generation with Vision Experts and Token Folding SynerGen-VL：利用视觉专家和Token Folding实现协同图像理解与生成 large language model multimodal
15 Do MLLMs Exhibit Human-like Perceptual Behaviors? HVSBench: A Benchmark for MLLM Alignment with Human Perceptual Behavior 提出HVSBench基准测试MLLM是否具备类人感知行为，揭示显著差距。 large language model multimodal
16 Lyra: An Efficient and Speech-Centric Framework for Omni-Cognition Lyra：一种高效且以语音为中心的全知认知框架 large language model multimodal
17 GenEx: Generating an Explorable World GenEx：通过生成式想象构建可探索的3D世界，提升具身智能体能力 embodied AI
18 TimeRefine: Temporal Grounding with Time Refining Video LLM TimeRefine：利用时间细化的视频LLM进行时序定位 TAMP
19 Vision-Language Models Generate More Homogeneous Stories for Phenotypically Black Individuals 视觉-语言模型对表型黑人生成更趋同的故事，揭示群体内部的同质性偏见 large language model
20 FD2-Net: Frequency-Driven Feature Decomposition Network for Infrared-Visible Object Detection 提出FD2-Net，通过频率驱动的特征分解实现红外-可见光图像目标检测性能提升。 multimodal
21 Towards Long-Horizon Vision-Language Navigation: Platform, Benchmark and Method 提出长程视觉-语言导航任务与基准，并设计多粒度动态记忆模型以提升导航性能。 VLN
22 Geo-LLaVA: A Large Multi-Modal Model for Solving Geometry Math Problems with Meta In-Context Learning 提出Geo-LLaVA，结合元上下文学习解决几何数学难题 large language model

🔬 支柱二：RL算法与架构 (RL & Architecture) (14 篇)

#题目一句话要点标签🔗⭐
23 Elevating Visual Perception in Multimodal LLMs with Visual Embedding Distillation VisPer-LM：通过视觉嵌入蒸馏提升多模态LLM的视觉感知能力 distillation embodied AI multimodal
24 VLMs meet UDA: Boosting Transferability of Open Vocabulary Segmentation with Unsupervised Domain Adaptation 提出UDA-FROVSS框架，结合VLM与UDA提升开放词汇语义分割的跨域迁移能力 distillation open-vocabulary open vocabulary
25 YingSound: Video-Guided Sound Effects Generation with Multi-modal Chain-of-Thought Controls YingSound：提出基于多模态CoT控制的视频引导音效生成方法，解决少样本场景下的高质量音效生成问题。 flow matching foundation model chain-of-thought ✅
26 Tuned Reverse Distillation: Enhancing Multimodal Industrial Anomaly Detection with Crossmodal Tuners 提出Tuned Reverse Distillation，通过跨模态调谐器增强多模态工业异常检测。 distillation multimodal ✅
27 Towards Robust and Fair Vision Learning in Open-World Environments 针对开放世界环境，论文提出提升视觉学习公平性和鲁棒性的方法 world model foundation model multimodal
28 Physics-Driven Autoregressive State Space Models for Medical Image Reconstruction 提出MambaRoll，一种物理驱动的自回归状态空间模型，用于高质量医学图像重建。 Mamba SSM state space model
29 USDRL: Unified Skeleton-Based Dense Representation Learning with Multi-Grained Feature Decorrelation 提出USDRL框架，通过多粒度特征解耦学习骨骼动作的稠密表征，提升动作识别、检索和检测性能。 representation learning contrastive learning ✅
30 Is Contrastive Distillation Enough for Learning Comprehensive 3D Representations? 提出CMCR框架，通过综合学习模态共享与特定特征，提升3D表征能力 representation learning distillation ✅
31 Selective Visual Prompting in Vision Mamba 针对Vision Mamba，提出选择性视觉提示（SVP）方法，提升下游任务微调性能。 Mamba state space model ✅
32 Dynamic Contrastive Knowledge Distillation for Efficient Image Restoration 提出动态对比知识蒸馏框架，提升图像复原任务中小型网络的性能。 contrastive learning distillation
33 Owl-1: Omni World Model for Consistent Long Video Generation 提出Owl-1：用于生成一致长视频的Omni World Model world model ✅
34 All You Need in Knowledge Distillation Is a Tailored Coordinate System 提出定制坐标系蒸馏方法，解决知识蒸馏对特定任务教师模型的依赖问题。 distillation
35 DomCLP: Domain-wise Contrastive Learning with Prototype Mixup for Unsupervised Domain Generalization 提出DomCLP，通过领域对比学习与原型混合解决无监督域泛化问题 contrastive learning ✅
36 Inference-Time Diffusion Model Distillation 提出Distillation++，通过推理时教师引导优化，提升扩散模型蒸馏性能。 distillation ✅

🔬 支柱三：空间感知与语义 (Perception & Semantics) (12 篇)

#题目一句话要点标签🔗⭐
37 Feat2GS: Probing Visual Foundation Models with Gaussian Splatting Feat2GS：利用高斯溅射探究视觉基础模型的3D感知能力 3DGS gaussian splatting splatting ✅
38 PBR-NeRF: Inverse Rendering with Physics-Based Neural Fields PBR-NeRF：利用物理渲染的神经场进行逆渲染，实现材质和光照的联合估计。 3D gaussian splatting gaussian splatting splatting ✅
39 FreeSplatter: Pose-free Gaussian Splatting for Sparse-view 3D Reconstruction FreeSplatter：无需相机位姿的稀疏视图高斯溅射三维重建 gaussian splatting splatting
40 GEAL: Generalizable 3D Affordance Learning with Cross-Modal Consistency GEAL：利用跨模态一致性提升3D可供性学习的泛化能力 gaussian splatting splatting affordance
41 eCARLA-scenes: A synthetically generated dataset for event-based optical flow prediction 提出eCARLA-scenes：用于事件相机光流预测的合成数据集，服务自动驾驶场景。 visual odometry optical flow
42 Stereo4D: Learning How Things Move in 3D from Internet Stereo Videos Stereo4D：从互联网立体视频中学习3D动态场景重建 depth estimation stereo depth scene reconstruction
43 LiftImage3D: Lifting Any Single Image to 3D Gaussians with Video Generation Priors LiftImage3D：利用视频生成先验将单张图像提升为3D高斯模型，解决单图3D重建难题。 3D gaussian splatting gaussian splatting splatting
44 SLAM3R: Real-Time Dense Scene Reconstruction from Monocular RGB Videos SLAM3R：一种基于单目RGB视频的实时稠密场景重建系统 scene reconstruction ✅
45 ResFlow: Fine-tuning Residual Optical Flow for Event-based High Temporal Resolution Motion Estimation ResFlow：微调残差光流，实现基于事件相机的高时间分辨率运动估计 optical flow
46 Labits: Layered Bidirectional Time Surfaces Representation for Event Camera-based Continuous Dense Trajectory Estimation 提出Labits：一种用于事件相机连续稠密轨迹估计的分层双向时间表面表示 optical flow TAMP
47 Cross-View Completion Models are Zero-shot Correspondence Estimators 提出基于跨视角补全模型的零样本对应关系估计方法 depth estimation ✅
48 Mojito: Motion Trajectory and Intensity Control for Video Generation Mojito：提出运动轨迹和强度可控的视频生成扩散模型 optical flow

🔬 支柱四：生成式动作 (Generative Motion) (2 篇)

#题目一句话要点标签🔗⭐
49 Motion Generation Review: Exploring Deep Learning for Lifelike Animation with Manifold 综述性论文：探索流形学习在人体动作生成中的深度学习应用 motion generation
50 Video Creation by Demonstration 提出δ-Diffusion，通过演示视频和上下文图像生成逼真且连贯的新视频 physically plausible foundation model ✅

🔬 支柱一：机器人控制 (Robot Control) (2 篇)

#题目一句话要点标签🔗⭐
51 Doe-1: Closed-Loop Autonomous Driving with Large World Model 提出Doe-1：基于大世界模型的闭环自动驾驶框架，实现统一感知、预测与规划 motion planning world model ✅
52 MS2Mesh-XR: Multi-modal Sketch-to-Mesh Generation in XR Environments MS2Mesh-XR：在XR环境中基于手绘草图和语音输入的多模态网格生成 manipulation ✅

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

#题目一句话要点标签🔗⭐
53 Identity-Preserving Pose-Guided Character Animation via Facial Landmarks Transformation 提出FLT方法，解决姿态引导的人物动画中面部一致性问题 character animation
54 Reversing the Damage: A QP-Aware Transformer-Diffusion Approach for 8K Video Restoration under Codec Compression 提出DiQP，一种基于Transformer-Diffusion的QP感知模型，用于恢复编解码压缩造成的8K视频质量损失。 spatiotemporal

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

#题目一句话要点标签🔗⭐
55 ContextHOI: Spatial Context Learning for Human-Object Interaction Detection ContextHOI：提出空间上下文学习框架，提升遮挡场景下人-物交互检测性能 human-object interaction HOI

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

#题目一句话要点标签🔗⭐
56 Weighted Poisson-disk Resampling on Large-Scale Point Clouds 提出加权泊松盘重采样方法，提升大规模点云处理的效率和几何一致性 geometric consistency

⬅️ 返回 cs.CV 首页 · 🏠 返回主页