NeurIPS 2021 accepted papers on 3D vision
59 of the 2,508 papers accepted at NeurIPS 2021 (2.4%) are on 3D vision, compared with 2.1% at NeurIPS 2020. They fall into 5 topics.
See NeurIPS 2021 in the interactive map
Submitting to NeurIPS? Checklist →
Working on something in this area? Describe your idea in scime atlas to see the closest papers by meaning and how crowded the spot has become.
Topics and papers
Face and shape · 20 papers
- Deep Marching Tetrahedra: a Hybrid Representation for High-Resolution 3D Shape Synthesis (105 citations)
- NeuS: Learning Neural Implicit Surfaces by Volume Rendering for Multi-view Reconstruction (85 citations)
- MetaAvatar: Learning Animatable Clothed Human Models from Few Depth Images (48 citations)
- ViSER: Video-Specific Surface Embeddings for Articulated 3D Shape Reconstruction (35 citations)
- SAPE: Spatially-Adaptive Progressive Encoding for Neural Optimization (23 citations)
- CorticalFlow: A Diffeomorphic Mesh Transformer Network for Cortical Surface Reconstruction (19 citations)
- OctField: Hierarchical Implicit Functions for 3D Modeling (13 citations)
- Volume Rendering of Neural Implicit Surfaces (10 citations)
- A Shading-Guided Generative Implicit Model for Shape-Accurate 3D-Aware Image Synthesis (8 citations)
- Differentiable rendering with perturbed optimizers (8 citations)
- To The Point: Correspondence-driven monocular 3D category reconstruction (8 citations)
- Coarse-to-fine Animal Pose and Shape Estimation (6 citations)
and 8 more in the interactive map.
Human pose estimation · 16 papers
- Direct Multi-view Multi-person 3D Pose Estimation (82 citations)
- Dynamics-regulated kinematic policy for egocentric pose estimation (38 citations)
- AP-10K: A Benchmark for Animal Pose Estimation in the Wild (36 citations)
- Multi-Person 3D Motion Prediction with Multi-Range Transformers (35 citations)
- The PAIR-R24M Dataset for Multi-animal 3D Pose Estimation (33 citations)
- ManiSkill: Generalizable Manipulation Skill Benchmark with Large-Scale Demonstrations (27 citations)
- Robust Pose Estimation in Crowded Scenes with Direct Pose-Level Inference (21 citations)
- Leveraging SE(3) Equivariance for Self-supervised Category-Level Object Pose Estimation from Point Clouds (17 citations)
- Garment4D: Garment Reconstruction from Point Cloud Sequences (10 citations)
- REMIPS: Physically Consistent 3D Reconstruction of Multiple Interacting People under Weak Supervision (10 citations)
- Aligning Silhouette Topology for Self-Adaptive 3D Human Pose Recovery (6 citations)
- 3DP3: 3D Scene Perception via Probabilistic Programming (5 citations)
and 4 more in the interactive map.
3D generation · 8 papers
- ATISS: Autoregressive Transformers for Indoor Scene Synthesis (47 citations)
- Habitat-Matterport 3D Dataset (HM3D): 1000 Large-scale 3D Environments for Embodied AI (36 citations)
- ARKitScenes: A Diverse Real-World Dataset For 3D Indoor Scene Understanding Using Mobile RGB-D Data (21 citations)
- Voxel-based 3D Detection and Reconstruction of Multiple Objects from a Single Image (15 citations)
- Panoptic 3D Scene Reconstruction From a Single RGB Image (8 citations)
- Dynamic Environments with Deformable Objects (4 citations)
- Residual Relaxation for Multi-view Representation Learning (3 citations)
- Object-Centric Representation Learning with Generative Spatial-Temporal Factorization (2 citations)
NeRF and radiance fields · 8 papers
- H-NeRF: Neural Radiance Fields for Rendering and Temporal Reconstruction of Humans in Motion (76 citations)
- Neural Human Performer: Learning Generalizable Radiance Fields for Human Performance Rendering (67 citations)
- NeRS: Neural Reflectance Surfaces for Sparse-view 3D Reconstruction in the Wild (51 citations)
- A-NeRF: Articulated Neural Radiance Fields for Learning Human Shape, Appearance, and Pose (19 citations)
- Neural-PIL: Neural Pre-Integrated Lighting for Reflectance Decomposition (19 citations)
- TöRF: Time-of-Flight Radiance Fields for Dynamic Scene View Synthesis (10 citations)
- Generative Occupancy Fields for 3D Surface-Aware Image Synthesis (4 citations)
- Neural Relightable Participating Media Rendering (2 citations)
Depth and stereo · 7 papers
- DROID-SLAM: Deep Visual SLAM for Monocular, Stereo, and RGB-D Cameras (231 citations)
- TransformerFusion: Monocular RGB Scene Reconstruction using Transformers (81 citations)
- Learning to dehaze with polarization (32 citations)
- Light Field Networks: Neural Scene Representations with Single-Evaluation Rendering (19 citations)
- DIB-R++: Learning to Predict Lighting and Material with a Hybrid Differentiable Renderer (9 citations)
- Hardware Design and Accurate Simulation of Structured-Light Scanning for Benchmarking of 3D Reconstruction Algorithms (4 citations)
- Dense Keypoints via Multiview Supervision
