Research map / 3D vision / Human pose estimation
3D-Aware Neural Body Fitting for Occlusion Robust 3D Human Pose Estimation
Yi Zhang, Pengliang Ji, Angtian Wang, Jieru Mei, Adam Kortylewski, Alan Yuille
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add NBV analysis-by-synthesis fitting on top of regression initialization (ablation) | 3DNBF · 3DPW-AdvOcc@80 | MPJPE: 171.4 -> 140.7 | Improved |
| use volumetric Neural Body Volumes instead of SMPL mesh representation (ablation) | 3DNBF · 3DPW-AdvOcc@80 | MPJPE: 146.6 -> 140.7 | Improved |
| add 3D-aware contrastive loss to feature extractor training (ablation) | 3DNBF · 3DPW-AdvOcc@80 | MPJPE: 167.4 -> 140.7 | Improved |
| use 4 pose-dependent features per kernel instead of 1 (pose-independent) (ablation) | 3DNBF · 3DPW-AdvOcc@80 | MPJPE: 189.6 -> 140.7 | Improved |
| increase pose-dependent features from 4 to 8 (ablation) | 3DNBF · 3DPW-AdvOcc@80 | MPJPE: 140.7 -> 161.2 | Hurt |
| replace regression baselines with 3DNBF on standard 3DPW | 3D human pose estimation · 3DPW | MPJPE: 81.4 -> 79.8 | Improved |
| use 3DNBF refinement instead of EFT refinement | HMR-EFT · 3DPW-AdvOcc@80 | MPJPE: 158.2 -> 150.2 | Improved |
| add SMPLify keypoint fitting to HMR-EFT on non-occluded 3DPW | HMR-EFT · 3DPW | MPJPE: 89.5 -> 106.2 | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- SPIN — SPIN: Structure-Preserving Inner Offset Network for Scene Text Recognition (AAAI 2021) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Occlusion-Aware Networks for 3D Human Pose Estimation in Video (ICCV 2019 · similarity 0.91)
- 3D Human Pose Estimation = 2D Pose Estimation + Matching (CVPR 2017 · similarity 0.90)
- 3D Human Pose Estimation From a Single Image via Distance Matrix Regression (CVPR 2017 · similarity 0.90)
- Human Body Model Fitting by Learned Gradient Descent (ECCV 2020 · similarity 0.89)
- $\text{Di}^2\text{Pose}$: Discrete Diffusion Model for Occluded 3D Human Pose Estimation (NeurIPS 2024 · similarity 0.88)
- Neural Localizer Fields for Continuous 3D Human Pose and Shape Estimation (NeurIPS 2024 · similarity 0.88)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
