cs.CV（2024-03-19）

📊 共 58 篇论文 | 🔗 13 篇有代码

🎯 兴趣领域导航

支柱三：空间感知与语义 (Perception & Semantics) (20 🔗5) 支柱九：具身大模型 (Embodied Foundation Models) (19 🔗6) 支柱六：视频提取与匹配 (Video Extraction) (6) 支柱二：RL算法与架构 (RL & Architecture) (5) 支柱一：机器人控制 (Robot Control) (4 🔗1) 支柱八：物理动画 (Physics-based Animation) (3 🔗1) 支柱五：交互与反应 (Interaction & Reaction) (1)

🔬 支柱三：空间感知与语义 (Perception & Semantics) (20 篇)

#题目一句话要点标签🔗⭐
1 HUGS: Holistic Urban 3D Scene Understanding via Gaussian Splatting 提出HUGS以解决城市场景整体理解问题 3D gaussian splatting gaussian splatting splatting
2 OV9D: Open-Vocabulary Category-Level 9D Object Pose and Size Estimation 提出OV9D以解决开放词汇类别级物体姿态与尺寸估计问题 open-vocabulary open vocabulary foundation model
3 DetToolChain: A New Prompting Paradigm to Unleash Detection Ability of MLLM 提出DetToolChain以提升多模态大语言模型的零-shot目标检测能力 open-vocabulary open vocabulary large language model
4 DecentNeRFs: Decentralized Neural Radiance Fields from Crowdsourced Images 提出DecentNeRF以解决集中式NeRF计算成本高的问题 NeRF neural radiance field
5 CLIP-VIS: Adapting CLIP for Open-Vocabulary Video Instance Segmentation 提出CLIP-VIS以解决开放词汇视频实例分割问题 open-vocabulary open vocabulary ✅
6 3D Semantic MapNet: Building Maps for Multi-Object Re-Identification in 3D 提出3D-Semantic MapNet以解决3D多物体重识别问题 semantic map egocentric zero-shot transfer
7 GaussianFlow: Splatting Gaussian Dynamics for 4D Content Creation 提出GaussianFlow以解决4D内容生成中的动态监督问题 gaussian splatting splatting optical flow ✅
8 IFFNeRF: Initialisation Free and Fast 6DoF pose estimation from a single image and a NeRF model 提出IFFNeRF以解决6DoF相机姿态估计问题 NeRF neural radiance field
9 Depth-guided NeRF Training via Earth Mover's Distance 提出基于地球移动距离的深度引导NeRF训练方法 NeRF neural radiance field
10 GVGEN: Text-to-3D Generation with Volumetric Representation 提出GVGEN以解决文本到3D生成效率低的问题 3D gaussian splatting 3D reconstruction gaussian splatting
11 Global-guided Focal Neural Radiance Field for Large-scale Scene Rendering 提出全球引导聚焦神经辐射场以解决大规模场景渲染模糊问题 NeRF neural radiance field ✅
12 RGBD GS-ICP SLAM 提出RGBD GS-ICP SLAM以解决高效定位与建图问题 3D gaussian splatting 3DGS gaussian splatting
13 PostoMETRO: Pose Token Enhanced Mesh Transformer for Robust 3D Human Mesh Recovery 提出PostoMETRO以解决极端场景下3D人类网格恢复问题 3D reconstruction human mesh recovery
14 VisionGPT: LLM-Assisted Real-Time Anomaly Detection for Safe Visual Navigation 提出VisionGPT以解决视觉导航中的异常检测问题 open-vocabulary open vocabulary large language model
15 FutureDepth: Learning to Predict the Future Improves Video Depth Estimation 提出FutureDepth以提升视频深度估计精度 depth estimation
16 Geometric Constraints in Deep Learning Frameworks: A Survey 提出几何约束以增强深度学习框架的场景理解能力 depth estimation scene understanding
17 TAPTR: Tracking Any Point with Transformers as Detection 提出TAPTR框架以解决任意点跟踪问题 optical flow
18 Learning Neural Volumetric Pose Features for Camera Localization 提出PoseMap以增强相机定位精度 NeRF
19 Instance-Warp: Saliency Guided Image Warping for Unsupervised Domain Adaptation 提出Instance-Warp以解决无监督领域适应中的背景干扰问题 scene understanding ✅
20 Vox-Fusion++: Voxel-based Neural Implicit Dense Tracking and Mapping with Multi-maps 提出Vox-Fusion++以解决大规模场景的稠密跟踪与映射问题 implicit representation ✅

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

#题目一句话要点标签🔗⭐
21 Contextual AD Narration with Interleaved Multimodal Sequence 提出Uni-AD以解决视觉障碍者的音频描述生成问题 foundation model multimodal ✅
22 Towards Multimodal In-Context Learning for Vision & Language Models 提出多轮课程学习方法以提升视觉语言模型的上下文学习能力 large language model multimodal
23 WaterVG: Waterway Visual Grounding based on Text-Guided Vision and mmWave Radar 提出WaterVG以解决无人水面车辆水道感知问题 visual grounding
24 Quantifying uncertainty in lung cancer segmentation with foundation models applied to mixed-domain datasets 提出快速评估肺癌分割模型不确定性的方法 foundation model
25 MEDBind: Unifying Language and Multimodal Medical Data Embeddings 提出MEDBind以解决多模态医学数据融合问题 multimodal
26 FUELVISION: A Multimodal Data Fusion and Multimodel Ensemble Algorithm for Wildfire Fuels Mapping 提出FUELVISION以解决野火燃料映射精度不足问题 multimodal
27 As Firm As Their Foundations: Can open-sourced foundation models be used to create adversarial examples for downstream tasks? 提出PRM攻击策略以揭示CLIP模型的脆弱性 foundation model
28 Chain-of-Spot: Interactive Reasoning Improves Large Vision-Language Models 提出Chain-of-Spot以解决视觉语言模型特征提取不足问题 multimodal instruction following ✅
29 mPLUG-DocOwl 1.5: Unified Structure Learning for OCR-free Document Understanding 提出统一结构学习以提升无OCR文档理解能力 large language model multimodal ✅
30 VisualCritic: Making LMMs Perceive Visual Quality Like Humans 提出VisualCritic以解决LMMs视觉质量感知不足问题 multimodal instruction following
31 Better Call SAL: Towards Learning to Segment Anything in Lidar 提出SAL方法以解决激光雷达目标分割问题 foundation model ✅
32 Bypassing LLM Watermarks with Color-Aware Substitutions 提出自颜色测试替代方法以解决LLM水印检测问题 large language model
33 Planner3D: LLM-enhanced graph prior meets 3D indoor scene explicit regularization 提出Planner3D以解决3D室内场景合成中的布局碰撞问题 large language model
34 RelationVLM: Making Large Vision-Language Models Understand Visual Relations 提出RelationVLM以解决视觉关系理解不足的问题 large language model
35 Discover and Mitigate Multiple Biased Subgroups in Image Classifiers 提出DIM方法以解决图像分类器中的多重偏见子群问题 foundation model ✅
36 Learning Cross-view Visual Geo-localization without Ground Truth 提出自监督学习框架以解决无标注图像的跨视角地理定位问题 foundation model
37 Task-Customized Mixture of Adapters for General Image Fusion 提出任务定制混合适配器以解决通用图像融合问题 foundation model ✅
38 SC-Diff: 3D Shape Completion with Latent Diffusion Models 提出SC-Diff以解决3D形状补全问题 multimodal
39 Boosting Transferability in Vision-Language Attacks via Diversification along the Intersection Region of Adversarial Trajectory 通过交叉区域多样化提升视觉语言攻击的可转移性 multimodal

🔬 支柱六：视频提取与匹配 (Video Extraction) (6 篇)

#题目一句话要点标签🔗⭐
40 Human Mesh Recovery from Arbitrary Multi-view Images 提出统一人类网格恢复框架以解决多视角图像问题 human mesh recovery HMR SMPL
41 WHAC: World-grounded Humans and Cameras 提出WHAC框架以解决单目视频中人类与相机轨迹估计问题 SMPL SMPL-X human motion
42 SceneScript: Reconstructing Scenes With An Autoregressive Structured Language Model 提出SceneScript以重建场景模型并解决传统方法不足 egocentric
43 HCPM: Hierarchical Candidates Pruning for Efficient Detector-Free Matching 提出HCPM以解决高效特征匹配问题 feature matching
44 Intention Action Anticipation Model with Guide-Feedback Loop Mechanism 提出HCR模型以解决人类意图预测问题 egocentric
45 Self-learning Canonical Space for Multi-view 3D Human Pose Estimation 提出自学习规范空间以解决多视角3D人体姿态估计问题 SMPL

🔬 支柱二：RL算法与架构 (RL & Architecture) (5 篇)

#题目一句话要点标签🔗⭐
46 TexDreamer: Towards Zero-Shot High-Fidelity 3D Human Texture Generation 提出TexDreamer以解决3D人类纹理生成的挑战 dreamer multimodal
47 Emotic Masked Autoencoder with Attention Fusion for Facial Expression Recognition 提出情感掩码自编码器与注意力融合以提升面部表情识别 masked autoencoder MAE
48 HYDRA: A Hyper Agent for Dynamic Compositional Visual Reasoning 提出HYDRA框架以解决动态组合视觉推理问题 reinforcement learning large language model
49 AnimateDiff-Lightning: Cross-Model Diffusion Distillation 提出AnimateDiff-Lightning以实现快速视频生成 distillation
50 UniBind: LLM-Augmented Unified and Balanced Representation Space to Bind Them All 提出UniBind以解决多模态表示空间不平衡问题 contrastive learning large language model

🔬 支柱一：机器人控制 (Robot Control) (4 篇)

#题目一句话要点标签🔗⭐
51 AnySkill: Learning Open-Vocabulary Physical Skill for Interactive Agents 提出AnySkill以解决开放词汇物理技能学习问题 humanoid imitation learning reward shaping
52 When Do We Not Need Larger Vision Models? 提出S$^2$方法以优化视觉模型规模问题 manipulation depth estimation multimodal ✅
53 Generalized Consistency Trajectory Models for Image Manipulation 提出广义一致性轨迹模型以解决图像处理问题 manipulation
54 Magic Fixup: Streamlining Photo Editing by Watching Dynamic Videos 提出Magic Fixup以解决动态视频中图像编辑问题 manipulation

🔬 支柱八：物理动画 (Physics-based Animation) (3 篇)

#题目一句话要点标签🔗⭐
55 Multimodal Fusion Method with Spatiotemporal Sequences and Relationship Learning for Valence-Arousal Estimation 提出多模态融合方法以解决情感估计问题 spatiotemporal multimodal
56 A conditional latent autoregressive recurrent model for generation and forecasting of beam dynamics in particle accelerators 提出条件潜在自回归递归模型以解决粒子加速器束流动态问题 spatiotemporal
57 EAS-SNN: End-to-End Adaptive Sampling and Representation for Event-based Detection with Recurrent Spiking Neural Networks 提出EAS-SNN以解决事件摄像头自适应采样问题 spatiotemporal ✅

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

#题目一句话要点标签🔗⭐
58 ViTGaze: Gaze Following with Interaction Features in Vision Transformers 提出ViTGaze以解决注视跟随中的交互特征提取问题 human-scene interaction

⬅️ 返回 cs.CV 首页 · 🏠 返回主页