Research map / 3D vision / Human pose estimation
Adaptive Computationally Efficient Network for Monocular 3D Hand Pose Estimation
Zhipeng Fan, Jun Liu, Yao Wang
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 gate instead of always using fine encoder | Vanilla-LSTM-Fine-HG · STB | GFLOPs per frame (PCK20 98.7% -> 97.9%): 6.96 -> 1.56 | Improved |
| add LSTM temporal refinement | Fine-HG · STB | 3D PCK20: 96.3% -> 98.7% | Improved |
| add LSTM temporal refinement | Coarse-HG · STB | 3D PCK20: 85.1% -> 92.1% | Improved |
| add LSTM temporal refinement | Fine-HG · FPHA | 3D PCK20: 79.7% -> 83.9% | Improved |
| average coarse and fine features without gate | Vanilla-LSTM-Fine-HG · FPHA | 3D PCK20: 83.9% -> 83.1% | Hurt |
| use Gaussian kernel gate instead of linear Neural Gate (ablation) | ACE · STB | 3D PCK20: 0.981 -> 0.983 | Improved |
| use Gaussian kernel gate instead of linear Neural Gate (ablation) | ACE · STB | GFLOPs per frame: 2.54 -> 2.09 | Improved |
| use coarse features in gate instead of temporal context only (ablation) | ACE · STB | 3D PCK20: 0.977 -> 0.983 | Improved |
| lower expected fine-encoder rate gamma_g from 0.3 to 0.01 (ablation) | ACE · STB | 3D PCK20 and GFLOPs: 0.984 -> 0.977; 3.30 -> 1.37 GFLOPs | Mixed |
| lower gamma_g from 0.01 to 0.001 (ablation) | ACE · STB | 3D PCK20: 0.977 -> 0.955 | Hurt |
| raise test threshold lambda from 0.5 to 0.7 (ablation) | ACE · STB | 3D PCK20: 0.983 -> 0.943; 2.09 -> 0.88 GFLOPs | Hurt |
| raise test threshold lambda from 0.1 to 0.5 (ablation) | ACE · STB | 3D PCK20 and GFLOPs: 0.987 -> 0.983; 7.01 -> 2.09 GFLOPs | Mixed |
Closest papers by meaning
- Efficiently Creating 3D Training Data for Fine Hand Pose Estimation (CVPR 2016 · similarity 0.89)
- Two Heads Are Better than One: Image-Point Cloud Network for Depth-Based 3D Hand Pose Estimation (AAAI 2023 · similarity 0.89)
- Real-Time 3D Hand Pose Estimation with 3D Convolutional Neural Networks (TPAMI 2018 · similarity 0.89)
- HOPE-Net: A Graph-Based Model for Hand-Object Pose Estimation (CVPR 2020 · similarity 0.89)
- Prior-aware Dynamic Temporal Modeling Framework for Sequential 3D Hand Pose Estimation (ICCV 2025 · similarity 0.89)
- Recurrent 3D Hand Pose Estimation Using Cascaded Pose-Guided 3D Alignments (TPAMI 2022 · 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.
