Research map / 3D vision / Human pose estimation
HandR2N2: Iterative 3D Hand Pose Estimation Using a Residual Recurrent Neural Network
Wencan Cheng, Jong Hwan Ko
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 the novel attentive GRU cell (ablation) | HandR2N2 · NYU | mean joint error (mm): 8.68 mm -> 8.06 mm | Improved |
| add channel-wise GCN to evolve hidden states (ablation) | HandR2N2 · NYU | mean joint error (mm): 7.99 mm -> 7.36 mm | Improved |
| add residual mechanism for relative representation (ablation) | HandR2N2 · NYU | mean joint error (mm): 7.62 mm -> 7.36 mm | Improved |
| replace conventional GCN with channel-wise GCN (ablation) | HandR2N2 · NYU | mean joint error (mm): 8.14 mm -> 7.76 mm | Improved |
| replace conventional GRU with attentive reset-gate GRU (ablation) | HandR2N2 · NYU | mean joint error (mm): 8.59 mm -> 7.99 mm | Improved |
| increase inference iterations from 1 to 4 (ablation) | HandR2N2 · ICVL | mean joint error (mm): 6.51 -> 5.70 | Improved |
| increase inference iterations from 4 to 5 (ablation) | HandR2N2 · ICVL | mean joint error (mm): 5.70 -> 5.71 | No clear effect |
| train with Itr >= 5 instead of 3 (ablation) | HandR2N2 · NYU | mean joint error (mm) | Hurt |
| train with Itr >= 3 and run more inference iterations than training (ablation) | HandR2N2 · NYU | mean joint error (mm): minimal error 7.27 mm after 5 inference iterations (Itr=3) | Improved |
| increase number of resampled points K beyond 48 (ablation) | HandR2N2 · NYU | mean joint error (mm) | No clear effect |
Closest papers by meaning
- Recurrent 3D Hand Pose Estimation Using Cascaded Pose-Guided 3D Alignments (TPAMI 2022 · similarity 0.91)
- Generalized Feedback Loop for Joint Hand-Object Pose Estimation (TPAMI 2019 · similarity 0.90)
- HandFoldingNet: A 3D Hand Pose Estimation Network Using Multiscale-Feature Guided Folding of a 2D Hand Skeleton (ICCV 2021 · similarity 0.89)
- Learning to Estimate 3D Hand Pose From Single RGB Images (ICCV 2017 · similarity 0.89)
- Real-Time 3D Hand Pose Estimation with 3D Convolutional Neural Networks (TPAMI 2018 · similarity 0.89)
- Hand PointNet: 3D Hand Pose Estimation Using Point Sets (CVPR 2018 · 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.
