Research map / 3D vision / Human pose estimation
AWR: Adaptive Weighting Regression for 3D Hand Pose Estimation
Weiting Huang, Pengfei Ren, Jingyu Wang, Qi Qi, Haifeng Sun
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 adaptive aggregation with joint-coordinate supervision to dense supervision (ablation) | AWR with 3D offsets · NYU | Mean Error (mm): 7.87 -> 7.48 | Improved |
| Use adaptive aggregation and joint supervision instead of direct coordinate regression | NYU ResNet50 · NYU | Mean Error (mm): 8.82 -> 7.75 | Improved |
| Use 3D offsets instead of 2D offsets with depth offset maps (ablation) | AWR with joint and dense supervision · NYU | Mean Error (mm): 7.58 -> 7.48 | Improved |
| Use depth offset maps rather than absolute depth maps (ablation) | AWR with 2D offsets · NYU | Mean Error (mm): 7.86 -> 7.58 | Improved |
| Increase backbone depth from ResNet50 to ResNet101 (ablation) | AWR · NYU | Mean Error (mm): 7.48 -> 7.37 | Improved |
| Replace ResNet50 with a two-stage Hourglass (ablation) | AWR · NYU | Mean Error (mm): 7.48 -> 7.43 | Improved |
| Add joint-coordinate supervision to dense-only training (ablation) | AWR with depth-image input · NYU | Mean Error (mm): 7.87 -> 7.48 | Improved |
| Add joint-coordinate supervision to dense-only training (ablation) | AWR with voxel input · NYU | Mean Error (mm): 11.40 -> 8.60 | Improved |
| Add joint-coordinate supervision to dense-only training (ablation) | AWR with point-cloud input · NYU | Mean Error (mm): 10.6 -> 10.52 | Improved |
| Increase the 3D-offset kernel size (ablation) | AWR with both supervisions · NYU | Mean Error (mm): 7.49mm; 7.48mm; 7.46mm; 7.45mm | No clear effect |
| Increase ResNet backbone depth from ResNet18 to ResNet50 (ablation) | Ours · HANDS 2017 | AVG mean error (mm): 7.84 -> 7.48 | Improved |
Closest papers by meaning
- Dense 3D Regression for Hand Pose Estimation (CVPR 2018 · similarity 0.88)
- 3D Hand Pose Estimation via Articulated Anchor-to-Joint 3D Local Regressors (TPAMI 2025 · similarity 0.88)
- Latent Regression Forest: Structured Estimation of 3D Hand Poses (TPAMI 2016 · similarity 0.87)
- Efficiently Creating 3D Training Data for Fine Hand Pose Estimation (CVPR 2016 · similarity 0.87)
- Self-Supervised 3D Hand Pose Estimation Through Training by Fitting (CVPR 2019 · similarity 0.86)
- Adaptive Computationally Efficient Network for Monocular 3D Hand Pose Estimation (ECCV 2020 · similarity 0.86)
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.
