Research map / 3D vision / Human pose estimation
Weakly-supervised 3D Hand Pose Estimation from Monocular RGB Images
Yujun Cai, Liuhao Ge, Jianfei Cai, Junsong Yuan
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 |
|---|---|---|---|
| fine-tune with 2D labels of real images (ablation) | RHD-pretrained network · STB (weakly-supervised) | 3D PCK AUC: 0.667 -> 0.807 | Improved |
| add depth regularizer to 2D fine-tuning (ablation) | weakly-supervised network · STB (weakly-supervised) | 3D PCK AUC: 0.807 -> 0.889 | Improved |
| add depth regularizer to 2D fine-tuning (ablation) | weakly-supervised network · STB (Fig. 5 legend) | 3D PCK AUC: 0.372 -> 0.508 | Improved |
| add depth regularizer to fully-supervised training (ablation) | fully-supervised network · RHD | 3D PCK AUC: 0.856 -> 0.887 | Improved |
| add depth regularizer to fully-supervised training (ablation) | fully-supervised network · STB | 3D PCK AUC: 0.989 -> 0.990 | No clear effect |
| add depth regularizer to fully-supervised training (ablation) | fully-supervised network · STB (Fig. 5 legend) | 3D PCK AUC: 0.714 -> 0.722 | Improved |
| add depth regularizer to fully-supervised training (ablation) | fully-supervised network · STB-BB (Fig. 7) | 3D PCK AUC: 0.993 -> 0.994 | No clear effect |
| add depth regularizer to fully-supervised training (ablation) | fully-supervised network · joint z-axis error (RHD, STB) | mean z error | Improved |
| use wrist-relative instead of palm-relative root on STB | fully-supervised network · STB | mean error: reducing the mean error by about 0.6mm | Hurt |
| full fully-supervised method vs Zimmermann et al. | fully-supervised network · RHD | 3D PCK AUC: 0.675 -> 0.887 | Improved |
Closest papers by meaning
- 3D Hand Pose Estimation Using Synthetic Data and Weakly Labeled RGB Images (TPAMI 2020 · similarity 0.98)
- Aligning Latent Spaces for 3D Hand Pose Estimation (ICCV 2019 · similarity 0.93)
- 3D Hand Shape and Pose Estimation From a Single RGB Image (CVPR 2019 · similarity 0.93)
- Weakly Supervised 3D Hand Pose Estimation via Biomechanical Constraints (ECCV 2020 · similarity 0.92)
- Learning to Estimate 3D Hand Pose From Single RGB Images (ICCV 2017 · similarity 0.91)
- Keypoint Fusion for RGB-D Based 3D Hand Pose Estimation (AAAI 2024 · similarity 0.89)
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.
