cs.CV（2023-11-28）

📊 共 66 篇论文 | 🔗 17 篇有代码

🎯 兴趣领域导航

支柱三：空间感知与语义 (Perception & Semantics) (19 🔗5) 支柱九：具身大模型 (Embodied Foundation Models) (19 🔗8) 支柱二：RL算法与架构 (RL & Architecture) (12 🔗4) 支柱四：生成式动作 (Generative Motion) (5) 支柱八：物理动画 (Physics-based Animation) (5) 支柱一：机器人控制 (Robot Control) (2) 支柱七：动作重定向 (Motion Retargeting) (2) 支柱六：视频提取与匹配 (Video Extraction) (2)

🔬 支柱三：空间感知与语义 (Perception & Semantics) (19 篇)

#题目一句话要点标签🔗⭐
1 Multi-Scale 3D Gaussian Splatting for Anti-Aliased Rendering 提出多尺度3D高斯点云以解决抗锯齿渲染问题 3D gaussian splatting 3D reconstruction gaussian splatting ✅
2 SCALAR-NeRF: SCAlable LARge-scale Neural Radiance Fields for Scene Reconstruction 提出SCALAR-NeRF以解决大规模场景重建问题 NeRF neural radiance field scene reconstruction ✅
3 UC-NeRF: Neural Radiance Field for Under-Calibrated Multi-view Cameras in Autonomous Driving 提出UC-NeRF以解决多视角相机系统的标定不足问题 depth estimation NeRF neural radiance field
4 Human Gaussian Splatting: Real-time Rendering of Animatable Avatars 提出基于3D高斯点云的人体模型以解决实时渲染问题 3D gaussian splatting gaussian splatting splatting
5 Active Open-Vocabulary Recognition: Let Intelligent Moving Mitigate CLIP Limitations 提出主动开放词汇识别以解决CLIP局限性问题 open-vocabulary open vocabulary embodied AI
6 REF$^2$-NeRF: Reflection and Refraction aware Neural Radiance Field 提出REF²-NeRF以解决玻璃物体场景中的反射与折射建模问题 3D reconstruction NeRF neural radiance field
7 Continuous Pose for Monocular Cameras in Neural Implicit Representation 提出连续优化单目相机姿态的方法以提升视觉任务性能 NeRF implicit representation motion representation
8 Emergent Open-Vocabulary Semantic Segmentation from Off-the-shelf Vision-Language Models 提出PnP-OVSS以解决开放词汇语义分割问题 open-vocabulary open vocabulary ✅
9 Rethinking Directional Integration in Neural Radiance Fields 提出神经辐射场的方向集成改进以提升视角依赖效果 3D reconstruction NeRF neural radiance field
10 Self-Supervised Motion Magnification by Backpropagating Through Optical Flow 提出自监督运动放大方法以解决微小运动检测问题 optical flow
11 DGNR: Density-Guided Neural Point Rendering of Large Driving Scenes 提出DGNR以解决大规模驾驶场景渲染效率与质量问题 depth estimation NeRF neural radiance field
12 StreamFlow: Streamlined Multi-Frame Optical Flow Estimation for Video Sequences 提出StreamFlow以解决视频序列中的光流估计问题 optical flow ✅
13 Panoptic Video Scene Graph Generation 提出全景视频场景图生成以解决视频理解中的局限性 scene understanding egocentric
14 SplitNeRF: Split Sum Approximation Neural Field for Joint Geometry, Illumination, and Material Estimation 提出SplitNeRF以解决真实物体几何、材质和光照估计问题 NeRF neural radiance field
15 The Sky's the Limit: Re-lightable Outdoor Scenes via a Sky-pixel Constrained Illumination Prior and Outside-In Visibility 提出基于天空像素约束的照明先验以解决户外场景重光照问题 NeRF ✅
16 HAtt-Flow: Hierarchical Attention-Flow Mechanism for Group Activity Scene Graph Generation in Videos 提出HAtt-Flow机制以解决视频中群体活动场景图生成问题 scene understanding
17 A Unified Approach for Text- and Image-guided 4D Scene Generation 提出Dream-in-4D以解决文本引导的4D场景生成问题 neural radiance field
18 SARA: Controllable Makeup Transfer with Spatial Alignment and Region-Adaptive Normalization 提出SARA方法以解决化妆风格转移中的空间错位问题 semantic map
19 Photo-SLAM: Real-time Simultaneous Localization and Photorealistic Mapping for Monocular, Stereo, and RGB-D Cameras 提出Photo-SLAM以解决便携设备上SLAM系统的资源消耗问题 implicit representation

🔬 支柱九：具身大模型 (Embodied Foundation Models) (19 篇)

#题目一句话要点标签🔗⭐
20 SEED-Bench-2: Benchmarking Multimodal Large Language Models 提出SEED-Bench-2以评估多模态大语言模型的层次能力 large language model multimodal ✅
21 Ranni: Taming Text-to-Image Diffusion for Accurate Instruction Following 提出Ranni以解决复杂文本指令下的图像生成问题 large language model instruction following ✅
22 Reason out Your Layout: Evoking the Layout Master from Large Language Models for Text-to-Image Synthesis 提出利用大语言模型生成布局以提升文本到图像合成质量 large language model chain-of-thought
23 UniIR: Training and Benchmarking Universal Multimodal Information Retrievers 提出UniIR以解决多模态信息检索的多样化需求 multimodal
24 Zero-shot Referring Expression Comprehension via Structural Similarity Between Images and Captions 提出基于结构相似性的零-shot指代表达理解方法 VLA foundation model visual grounding ✅
25 LLaMA-VID: An Image is Worth 2 Tokens in Large Language Models 提出LLaMA-VID以解决长视频处理中的token生成问题 large language model ✅
26 Large Language Models Meet Computer Vision: A Brief Survey 探讨大语言模型与计算机视觉的结合以推动AI进步 large language model
27 Efficient Multimodal Diffusion Models Using Joint Data Infilling with Partially Shared U-Net 提出部分共享U-Net以解决多模态扩散模型效率问题 multimodal
28 Unsupervised Multimodal Deepfake Detection Using Intra- and Cross-Modal Inconsistencies 提出无监督多模态深伪检测方法以解决现有方法的局限性 multimodal
29 COLE: A Hierarchical Generation Framework for Multi-Layered and Editable Graphic Design 提出COLE框架以解决多层次可编辑图形设计问题 large language model multimodal
30 Large Model Based Referring Camouflaged Object Detection 提出基于大模型的多层知识引导方法以解决伪装目标检测问题 large language model multimodal
31 MobileCLIP: Fast Image-Text Models through Multi-Modal Reinforced Training 提出MobileCLIP以解决移动设备上图像-文本模型的效率问题 foundation model ✅
32 Eliciting In-Context Learning in Vision-Language Models for Videos Through Curated Data Distributional Properties 提出EILEV以解决视频理解中的上下文学习问题 large language model ✅
33 Telling Left from Right: Identifying Geometry-Aware Semantic Correspondence 提出几何感知语义对应方法以解决现有模型不足问题 foundation model ✅
34 MVBench: A Comprehensive Multi-modal Video Understanding Benchmark 提出MVBench以解决视频理解中的时序理解问题 large language model ✅
35 Parameter Efficient Fine-tuning via Cross Block Orchestration for Segment Anything Model 提出跨块协同机制以解决图像分割中的参数高效微调问题 foundation model
36 MotionZero:Exploiting Motion Priors for Zero-shot Text-to-Video Generation 提出MotionZero以解决零-shot文本到视频生成中的运动先验问题 large language model
37 Agents meet OKR: An Object and Key Results Driven Agent System with Hierarchical Self-Collaboration and Self-Evaluation 提出OKR-Agent以提升大语言模型的任务解决能力 large language model
38 TextDiffuser-2: Unleashing the Power of Language Models for Text Rendering 提出TextDiffuser-2以解决文本渲染中的灵活性和多样性问题 large language model

🔬 支柱二：RL算法与架构 (RL & Architecture) (12 篇)

#题目一句话要点标签🔗⭐
39 HumanGaussian: Text-Driven 3D Human Generation with Gaussian Splatting 提出HumanGaussian以解决文本驱动的3D人类生成问题 distillation 3D gaussian splatting gaussian splatting ✅
40 Beyond Hallucinations: Enhancing LVLMs through Hallucination-Aware Direct Preference Optimization 提出HA-DPO以解决多模态大语言模型的幻觉问题 preference learning DPO direct preference optimization ✅
41 LightGaussian: Unbounded 3D Gaussian Compression with 15x Reduction and 200+ FPS 提出LightGaussian以解决3D高斯表示存储效率低的问题 distillation 3D gaussian splatting gaussian splatting
42 LLaFS: When Large Language Models Meet Few-Shot Segmentation 提出LLaFS以解决少样本分割中的信息不足问题 curriculum learning large language model
43 A Mixture of Exemplars Approach for Efficient Out-of-Distribution Detection with Foundation Models 提出混合示例方法以高效检测分布外数据 representation learning foundation model
44 Adversarial Diffusion Distillation 提出对抗性扩散蒸馏方法以实现快速高质量图像合成 distillation foundation model ✅
45 Straighter Flow Matching via a Diffusion-Based Coupling Prior 提出StraightFM以解决流匹配中的轨迹直线化问题 flow matching multimodal
46 CLAP: Isolating Content from Style through Contrastive Learning with Augmented Prompts 提出CLAP以解决视觉语言模型内容与风格混合问题 contrastive learning multimodal
47 F4D: Factorized 4D Convolutional Neural Network for Efficient Video-level Representation Learning 提出F4D以解决视频级表示学习中的长时序建模问题 representation learning spatiotemporal
48 RichDreamer: A Generalizable Normal-Depth Diffusion Model for Detail Richness in Text-to-3D 提出RichDreamer以解决文本到3D生成中的细节丰富性问题 dreamer distillation ✅
49 Embodied Multi-Modal Agent trained by an LLM from a Parallel TextWorld 提出EMMA以解决多模态智能体在视觉世界中的适应问题 imitation learning DPO large language model
50 DiffusionTalker: Personalization and Acceleration for Speech-Driven 3D Face Diffuser 提出DiffusionTalker以解决个性化与加速生成3D面部动画问题 contrastive learning distillation

🔬 支柱四：生成式动作 (Generative Motion) (5 篇)

#题目一句话要点标签🔗⭐
51 A Unified Framework for Multimodal, Multi-Part Human Motion Synthesis 提出统一框架以解决多模态、多部位人类动作合成问题 motion synthesis motion generation human motion
52 TLControl: Trajectory and Language Control for Human Motion Synthesis 提出TLControl以解决人类运动合成中的语言与轨迹控制问题 motion synthesis VQ-VAE human motion
53 ReMoS: 3D Motion-Conditioned Reaction Synthesis for Two-Person Interactions 提出ReMoS以解决多人体交互中的运动合成问题 motion synthesis two-person interaction reaction synthesis
54 AvatarGPT: All-in-One Framework for Motion Understanding, Planning, Generation and Beyond 提出AvatarGPT以解决人类动作理解与生成问题 motion synthesis human motion large language model
55 Space-Time Diffusion Features for Zero-Shot Text-Driven Motion Transfer 提出基于时空扩散特征的零-shot文本驱动运动转移方法 text-driven motion

🔬 支柱八：物理动画 (Physics-based Animation) (5 篇)

#题目一句话要点标签🔗⭐
56 Animate Anyone: Consistent and Controllable Image-to-Video Synthesis for Character Animation 提出基于扩散模型的角色动画生成框架以解决视频一致性问题 character animation
57 Pattern retrieval of traffic congestion using graph-based associations of traffic domain-specific features 提出基于图的关联方法以解决交通拥堵模式检索问题 spatiotemporal
58 Image segmentation with traveling waves in an exactly solvable recurrent neural network 提出基于可解递归神经网络的图像分割方法 spatiotemporal
59 Riemannian Self-Attention Mechanism for SPD Networks 提出Riemannian自注意力机制以改善SPD网络性能 spatiotemporal
60 Augmenting x-ray single particle imaging reconstruction with self-supervised machine learning 提出自监督机器学习方法以增强X射线单粒子成像重建 PULSE

🔬 支柱一：机器人控制 (Robot Control) (2 篇)

#题目一句话要点标签🔗⭐
61 Point'n Move: Interactive Scene Object Manipulation on Gaussian Splatting Radiance Fields 提出Point'n Move以实现交互式场景物体操控 manipulation gaussian splatting splatting
62 HandyPriors: Physically Consistent Perception of Hand-Object Interactions with Differentiable Priors 提出HandyPriors以解决手-物体交互的感知问题 manipulation penetration human-object interaction

🔬 支柱七：动作重定向 (Motion Retargeting) (2 篇)

#题目一句话要点标签🔗⭐
63 Unlocking Spatial Comprehension in Text-to-Image Diffusion Models 提出CompFuser以解决文本到图像生成模型的空间理解问题 spatial relationship large language model
64 Neural Texture Puppeteer: A Framework for Neural Geometry and Texture Rendering of Articulated Shapes, Enabling Re-Identification at Interactive Speed 提出神经纹理木偶框架以实现交互速度的重识别 spatial relationship

🔬 支柱六：视频提取与匹配 (Video Extraction) (2 篇)

#题目一句话要点标签🔗⭐
65 Egocentric Whole-Body Motion Capture with FisheyeViT and Diffusion-Based Motion Refinement 提出基于FisheyeViT的单目全身动作捕捉方法以解决自遮挡问题 egocentric
66 Exo2EgoDVC: Dense Video Captioning of Egocentric Procedural Activities Using Web Instructional Videos 提出Exo2EgoDVC以解决视频字幕跨视角转移问题 egocentric

⬅️ 返回 cs.CV 首页 · 🏠 返回主页