Research map / 3D vision / Human pose estimation
Robust 3D Hand Pose Estimation in Single Depth Images: From Single-View CNN to Multi-View CNNs
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| use three-view CNN projections with fine heat-map fusion rather than single-view regression (ablation) | 3D hand pose estimation · Self-comparison dataset | mean joint error | Helped |
| use three-view CNN projections with fine heat-map fusion rather than single-view regression (ablation) | 3D hand pose estimation · Self-comparison dataset | worst-case accuracy | Helped |
| use fine probabilistic fusion instead of averaging the three views' estimated coordinates (ablation) | Multi-view regression · Self-comparison dataset | mean joint error: about 13 mm | Helped |
| use fine probabilistic fusion instead of averaging the three views' estimated coordinates (ablation) | Multi-view regression · Self-comparison dataset at large error tolerances | worst-case accuracy | Hurt |
| use fine probabilistic fusion instead of averaging the three views' estimated coordinates (ablation) | Multi-view regression · Self-comparison dataset at small error tolerances | worst-case accuracy | Helped |
| use 35 principal components for the hand-pose constraint (ablation) | Multi-view fine fusion · Self-comparison dataset | worst-case accuracy | Helped |
| use multi-view heat-map regression rather than the compared single-view CNN method | 3D hand pose estimation · State-of-the-art comparison dataset | worst-case accuracy | Helped |
| use multi-view fine fusion rather than hierarchical regression | 3D hand pose estimation · State-of-the-art comparison dataset at 20mm and 30 mm tolerances | good-sample proportion: about 10%; 15% | Helped |
| use multi-view fine fusion rather than hierarchical regression | 3D hand pose estimation · State-of-the-art comparison dataset below 15mm tolerance | good-sample proportion | Hurt |
| use multi-view fine fusion rather than hierarchical regression | 3D hand pose estimation · State-of-the-art comparison dataset across yaw and pitch viewpoints | mean estimation error | Helped |
| train on one dataset and test without calibration, temporal information, or ground-truth initialization | Multi-view CNN · Cross-dataset test set (Avg) | average estimation error (mm): 28.6 -> 22.8; 23.6 -> 22.8; 26.1 -> 22.8 | Helped |
| use multi-view CNN regression without calibrated hand models or temporal information rather than ICP-PSO tracking | Cross-dataset hand pose estimation · Cross-dataset test set (Avg) | average estimation error (mm): 22.8 vs 15.3; 22.8 vs 9.2 | Hurt |
Closest papers by meaning
- Dense 3D Regression for Hand Pose Estimation (CVPR 2018 · similarity 0.94)
- Depth-Based 3D Hand Pose Estimation: From Current Achievements to Future Goals (CVPR 2018 · similarity 0.93)
- 3D Convolutional Neural Networks for Efficient and Robust Hand Pose Estimation From Single Depth Images (CVPR 2017 · similarity 0.92)
- DeepHand: Robust Hand Pose Estimation by Completing a Matrix Imputed With Deep Features (CVPR 2016 · similarity 0.92)
- Self-Supervised 3D Hand Pose Estimation Through Training by Fitting (CVPR 2019 · similarity 0.92)
- Keypoint Fusion for RGB-D Based 3D Hand Pose Estimation (AAAI 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.