Research map / 3D vision / Human pose estimation
Weakly Supervised 3D Hand Pose Estimation via Biomechanical Constraints
Adrian Spurr, Umar Iqbal, Pavlo Molchanov, Otmar Hilliges, Jan Kautz
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 biomechanical constraint losses (BMC) | RHD 3D + FH 2D weak supervision · FreiHAND val | 3D error (mm): 20.92 -> 13.78 | Improved |
| add BMC | RHD 3D + FH 2D weak supervision · FreiHAND val | depth Z error (mm): 20.02 -> 9.97 | Improved |
| add real 2D labels (naive weak supervision) | RHD-trained model · FreiHAND val | 3D error (mm): 30.82 -> 20.92 | Improved |
| add 2D labels only (no BMC) | RHD-trained model · FreiHAND val | depth Z error (mm): 20.02 -> 17.02 | No clear effect |
| add BMC to fully 3D supervised model | RHD+FH 3D · FreiHAND val | 3D error (mm): 8.78 -> 8.6 | Improved |
| add Z_root refinement (ablation) | FH fully supervised · FreiHAND val | mean EPE (mm): 11.20 -> 9.76 | Improved |
| add bone length loss L_BL (ablation) | 3D_RHD + 2D_FH · FreiHAND val | mean EPE (mm): 20.92 -> 17.58 | Improved |
| add root bone loss L_RB (ablation) | 3D_RHD + 2D_FH + L_BL · FreiHAND val | mean EPE (mm): 17.58 -> 15.48 | Improved |
| add angle loss L_A (ablation) | 3D_RHD + 2D_FH + L_BL + L_RB · FreiHAND val | mean EPE (mm): 15.48 -> 13.78 | Improved |
| model flexion and abduction limits as co-dependent (ablation) | BMC angle loss · FreiHAND val | mean EPE (mm): 15.57 -> 13.78 | Improved |
| use limits computed from the data instead of approximated from RHD (ablation) | BMC · FreiHAND val | mean EPE (mm): 16.14 -> 13.78 | Improved |
| model flexion/abduction angles as co-dependent (ablation) | BMC angle loss · HO-3D val | mean EPE (mm): 18.30 -> 18.50 | No clear effect |
Closest papers by meaning
- Weakly-supervised 3D Hand Pose Estimation from Monocular RGB Images (ECCV 2018 · similarity 0.92)
- Weakly-Supervised 3D Hand Reconstruction with Knowledge Prior and Uncertainty Guidance (ECCV 2024 · similarity 0.92)
- 3D Hand Pose Estimation Using Synthetic Data and Weakly Labeled RGB Images (TPAMI 2020 · similarity 0.91)
- Efficiently Creating 3D Training Data for Fine Hand Pose Estimation (CVPR 2016 · similarity 0.90)
- 3D Hand Shape and Pose From Images in the Wild (CVPR 2019 · similarity 0.90)
- Efficient Annotation and Learning for 3D Hand Pose Estimation: A Survey (IJCV 2023 · similarity 0.90)
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.
