CVPR 2023 accepted papers on 3D vision
491 of the 2,359 papers accepted at CVPR 2023 (20.8%) are on 3D vision, compared with 17.8% at CVPR 2022. They fall into 5 topics.
See CVPR 2023 in the interactive map
Working on something in this area? Describe your idea in scime atlas to see the closest accepted papers by meaning and how crowded the spot has become.
Topics and papers
Face and shape · 126 papers
- Neuralangelo: High-Fidelity Neural Surface Reconstruction (431 citations)
- SadTalker: Learning Realistic 3D Motion Coefficients for Stylized Audio-Driven Single Image Talking Face Animation (312 citations)
- SDFusion: Multimodal 3D Shape Completion, Reconstruction, and Generation (201 citations)
- ECON: Explicit Clothed Humans Optimized via Normal Integration (195 citations)
- CodeTalker: Speech-Driven 3D Facial Animation With Discrete Motion Prior (183 citations)
- RealFusion: 360deg Reconstruction of Any Object From a Single Image (178 citations)
- Instant Volumetric Head Avatars (174 citations)
- PointAvatar: Deformable Point-Based Head Avatars From Videos (167 citations)
- InstantAvatar: Learning Avatars From Monocular Video in 60 Seconds (137 citations)
- Vid2Avatar: 3D Avatar Reconstruction From Videos in the Wild via Self-Supervised Scene Decomposition (117 citations)
- Next3D: Generative Neural Texture Rasterization for 3D-Aware Head Avatars (97 citations)
- Diffusion-SDF: Text-To-Shape via Voxelized Diffusion (91 citations)
and 114 more in the interactive map.
Depth and stereo · 103 papers
- Iterative Geometry Encoding Volume for Stereo Matching (343 citations)
- Co-SLAM: Joint Coordinate and Sparse Parametric Encodings for Neural Real-Time SLAM (269 citations)
- Tensor4D: Efficient Neural 4D Decomposition for High-Fidelity Dynamic Reconstruction and Rendering (176 citations)
- DKM: Dense Kernelized Feature Matching for Geometry Estimation (172 citations)
- GeoMVSNet: Learning Multi-View Stereo With Geometry Perception (114 citations)
- iDisc: Internal Discretization for Monocular Depth Estimation (114 citations)
- Collaboration Helps Camera Overtake LiDAR in 3D Detection (103 citations)
- Accelerated Coordinate Encoding: Learning to Relocalize in Minutes Using RGB and Poses (97 citations)
- OrienterNet: Visual Localization in 2D Public Maps With Neural Matching (92 citations)
- Spring: A High-Resolution High-Detail Dataset and Benchmark for Scene Flow, Optical Flow and Stereo (71 citations)
- Deep Depth Estimation From Thermal Image (66 citations)
- Residual Degradation Learning Unfolding Framework With Mixing Priors Across Spectral and Spatial for Compressive Spectral Imaging (66 citations)
and 91 more in the interactive map.
NeRF and radiance fields · 94 papers
- K-Planes: Explicit Radiance Fields in Space, Time, and Appearance (545 citations)
- HexPlane: A Fast Representation for Dynamic Scenes (425 citations)
- FreeNeRF: Improving Few-Shot Neural Rendering With Free Frequency Regularization (307 citations)
- MobileNeRF: Exploiting the Polygon Rasterization Pipeline for Efficient Neural Field Rendering on Mobile Architectures (304 citations)
- ESLAM: Efficient Dense SLAM System Based on Hybrid Representation of Signed Distance Fields (241 citations)
- NoPe-NeRF: Optimising Neural Radiance Field With No Pose Prior (230 citations)
- Robust Dynamic Radiance Fields (146 citations)
- SPIn-NeRF: Multiview Segmentation and Perceptual Inpainting With Neural Radiance Fields (139 citations)
- TensoIR: Tensorial Inverse Rendering (137 citations)
- SPARF: Neural Radiance Fields From Sparse and Noisy Poses (135 citations)
- DiffRF: Rendering-Guided 3D Radiance Field Diffusion (127 citations)
- F2-NeRF: Fast Neural Radiance Field Training With Free Camera Trajectories (123 citations)
and 82 more in the interactive map.
Human pose estimation · 86 papers
- PoseFormerV2: Exploring Frequency Domain for Efficient and Robust 3D Human Pose Estimation (254 citations)
- 3D Human Pose Estimation With Spatio-Temporal Criss-Cross Attention (218 citations)
- DiffPose: Toward More Reliable 3D Pose Estimation (184 citations)
- Generating Holistic 3D Human Motion From Speech (160 citations)
- BEDLAM: A Synthetic Dataset of Bodies Exhibiting Detailed Lifelike Animated Motion (141 citations)
- TryOnDiffusion: A Tale of Two UNets (130 citations)
- One-Stage 3D Whole-Body Mesh Recovery With Component Aware Transformer (116 citations)
- UniDexGrasp: Universal Robotic Dexterous Grasping via Learning Diverse Proposal Generation and Goal-Conditioned Policy (113 citations)
- Avatars Grow Legs: Generating Smooth Human Motion From Sparse Tracking Inputs With Diffusion Model (99 citations)
- 3D Human Pose Estimation via Intuitive Physics (96 citations)
- Human Pose As Compositional Tokens (91 citations)
- Decoupling Human and Camera Motion From Videos in the Wild (81 citations)
and 74 more in the interactive map.
3D generation · 82 papers
- OpenScene: 3D Scene Understanding With Open Vocabularies (365 citations)
- Score Jacobian Chaining: Lifting Pretrained 2D Diffusion Models for 3D Generation (337 citations)
- Tri-Perspective View for Vision-Based 3D Semantic Occupancy Prediction (307 citations)
- VoxFormer: Sparse Voxel Transformer for Camera-Based 3D Semantic Scene Completion (244 citations)
- OmniObject3D: Large-Vocabulary 3D Object Dataset for Realistic Perception, Reconstruction and Generation (188 citations)
- Diffusion-Based Generation, Optimization, and Planning in 3D Scenes (184 citations)
- SparseFusion: Distilling View-Conditioned Diffusion for 3D Reconstruction (170 citations)
- 3D Neural Field Generation Using Triplane Diffusion (164 citations)
- ARCTIC: A Dataset for Dexterous Bimanual Hand-Object Manipulation (162 citations)
- BundleSDF: Neural 6-DoF Tracking and 3D Reconstruction of Unknown Objects (160 citations)
- UniSim: A Neural Closed-Loop Sensor Simulator (157 citations)
- Solving 3D Inverse Problems Using Pre-Trained 2D Diffusion Models (147 citations)
and 70 more in the interactive map.