Research map / 3D vision / Human pose estimation
Prior-aware Dynamic Temporal Modeling Framework for Sequential 3D Hand Pose Estimation
Pengfei Ren, Jingyu Wang, Haifeng Sun, Qi Qi, Xingyu Liu, Menghao Zhang, Lei Zhang, Jing Wang, Jianxin Liao
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 |
|---|---|---|---|
| insert joint-level dynamic TCN before attention (head) (ablation) | MotionBERT dual-branch · InterHand2.6M (ablation) | MPJPE (mm): 8.13 -> 8.03 | Improved |
| insert joint-level dynamic TCN after attention (tail) (ablation) | MotionBERT dual-branch · InterHand2.6M (ablation) | MPJPE (mm): 8.13 -> 7.70 | Improved |
| replace temporal attention with dynamic TCN (ablation) | MotionBERT dual-branch · InterHand2.6M (ablation) | MPJPE (mm): 8.13 -> 8.25 | Hurt |
| use tri-branch instead of dual-branch dynamic block (ablation) | Dynamic dual-branch · InterHand2.6M (ablation) | MPJPE (mm): 7.64 -> 7.59 | Improved |
| insert dynamic TCN after fusion of two branches (ablation) | Dynamic dual-branch · InterHand2.6M (ablation) | MPJPE (mm): 7.64 -> 7.78 | Hurt |
| add serial global temporal block with attention and hand-level dynamic TCN (ablation) | Dynamic dual-branch · InterHand2.6M (ablation) | MPJPE (mm): 7.64 -> 7.42 | Improved |
| replace joint-level dynamic TCN with hand-level dynamic TCN (ablation) | Dynamic dual-branch · InterHand2.6M (ablation) | MPJPE (mm): 7.64 -> 7.85 | Hurt |
| add hand prior memory with 8 tokens (ablation) | Dynamic dual-branch with global block · InterHand2.6M (ablation) | MPJPE (mm): 7.42 -> 7.21 | Improved |
| increase memory tokens from 8 to 16 (ablation) | Hand prior memory · InterHand2.6M (ablation) | MPJPE (mm): 7.21 -> 7.28 | Hurt |
| use ViT-style single token instead of memory (ablation) | Hand prior memory · InterHand2.6M (ablation) | MPJPE (mm): 7.21 -> 7.40 | Hurt |
| supervise only the final updated memory (ablation) | Hand prior memory · InterHand2.6M (ablation) | MPJPE (mm): 7.35 -> 7.21 | Improved |
| add the temporal refiner to InterWild | InterWild · InterHand2.6M (30 FPS) | MPJPE (mm): 10.32 -> 7.21 | Improved |
Closest papers by meaning
- SeqHAND: RGB-Sequence-Based 3D Hand Pose and Shape Estimation (ECCV 2020 · similarity 0.91)
- Hand-Transformer: Non-Autoregressive Structured Modeling for 3D Hand Pose Estimation (ECCV 2020 · similarity 0.90)
- Deformer: Dynamic Fusion Transformer for Robust Hand Pose Estimation (ICCV 2023 · similarity 0.90)
- ComPose: When to Trust Hands for Object Pose Tracking (NeurIPS 2026 · similarity 0.89)
- Pose-Guided Temporal Enhancement for Robust Low-Resolution Hand Reconstruction (CVPR 2025 · similarity 0.89)
- Cross-Constrained Progressive Inference for 3D Hand Pose Estimation with Dynamic Observer-Decision-Adjuster Networks (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.
