Research map / 3D vision / Human pose estimation
Spatial-Temporal Parallel Transformer for Arm-Hand Dynamic Estimation
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 arm-hand key-points jointly instead of hand key-points alone (ablation) | CNN gesture and dynamics estimator · Motion Capture Dataset test set | MPJPE (hands): 0.0300 -> 0.0130 | Helped |
| Add smooth loss alone (ablation) | CNN ah2ah · Motion Capture Dataset test set | MPJPE (overall): 0.0517 -> 0.0519 | No clear effect |
| Add smooth loss alone (ablation) | CNN ah2ah · Motion Capture Dataset test set | MPJRE (arms): 0.0604 -> 0.0603 | No clear effect |
| Add FK loss (ablation) | CNN ah2ah · Motion Capture Dataset test set | MPJPE (overall): 0.0517 -> 0.0413 | Helped |
| Add FK loss (ablation) | CNN ah2ah · Motion Capture Dataset test set | MPJRE (arms): 0.0604 -> 0.0579 | Helped |
| Add smooth loss together with FK loss (ablation) | CNN ah2ah · Motion Capture Dataset test set | MPJPE (overall): 0.0413 -> 0.0408 | Helped |
| Replace CNN with a temporal transformer (ablation) | Arm-hand dynamics estimator · Motion Capture Dataset test set | MPJPE (overall): 0.0408 -> 0.0316 | Helped |
| Add a spatial transformer to the temporal transformer (ablation) | AHMT · Motion Capture Dataset test set | MPJPE (overall): 0.0316 -> 0.0274 | Helped |
| Add a spatial transformer to the temporal transformer (ablation) | AHMT · Motion Capture Dataset test set | MPJRE (arms): 0.0445 -> 0.0375 | Helped |
| Use PAHMT instead of Body2Hands* | Hand pose estimation · BH Dataset | MPJPE (hands): 0.0346 -> 0.0281 | Helped |
| Use PAHMT instead of ExPose | Full body motion capture · Rendered Full Body Motion Dataset | MPJPE (overall): 0.1530 -> 0.1375 | Helped |
| Use PAHMT instead of ExPose | Full body motion capture · Rendered Full Body Motion Dataset | MPJRE (arms): 0.2984 -> 0.1614 | Helped |
Closest papers by meaning
- Dyn-HaMR: Recovering 4D Interacting Hand Motion from a Dynamic Camera (CVPR 2025 · similarity 0.87)
- Hand-Centric Motion Refinement for 3D Hand-Object Interaction via Hierarchical Spatial-Temporal Modeling (AAAI 2024 · similarity 0.86)
- StableHand: Quality-Aware Flow Matching for World-Space Dual-Hand Motion Estimation from Egocentric Video (NeurIPS 2026 · similarity 0.86)
- ComPose: When to Trust Hands for Object Pose Tracking (NeurIPS 2026 · similarity 0.86)
- HaWoR: World-Space Hand Motion Reconstruction from Egocentric Videos (CVPR 2025 · similarity 0.85)
- HOLD: Category-agnostic 3D Reconstruction of Interacting Hands and Objects from Video (CVPR 2024 · similarity 0.85)
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.