Research map / Video / Action recognition
Hierarchical Temporal Transformer for 3D Hand Pose Estimation and Action Recognition From Egocentric RGB Videos
Yilin Wen, Hao Pan, Lei Yang, Jia Pan, Taku Komura, Wenping 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 |
|---|---|---|---|
| Use a short temporal pose window (t=16) instead of image-only pose estimation (ablation) | HTT pose block · FPHA | AUC(0-80): 0.776 -> 0.810 | Improved |
| Use a short temporal pose window (t=16) instead of image-only pose estimation (ablation) | HTT pose block · H2O, left hand | MEPE: 40.12 -> 35.02 | Improved |
| Extend pose context from t=16 to t=128 (ablation) | HTT pose block · FPHA | AUC(0-80): 0.810 -> 0.800 | Hurt |
| Extend pose context from t=16 to t=128 (ablation) | HTT pose block · H2O, left hand | MEPE: 35.02 -> 36.41 | Hurt |
| Extend action context from T=16 to T=128 (ablation) | HTT action block · H2O | Classification accuracy: 74.38 -> 86.36 | Improved |
| Extend action context from T=16 to T=128 (ablation) | HTT action block · FPHA | Classification accuracy: 90.96 -> 94.09 | Improved |
| Cascade pose and action blocks instead of running them in parallel (ablation) | HTT action recognition · FPHA | Classification accuracy: 93.22 -> 94.09 | Improved |
| Cascade pose and action blocks instead of running them in parallel (ablation) | HTT action recognition · H2O | Classification accuracy: 80.17 -> 86.36 | Improved |
| Add object label to the action-block input (ablation) | HTT action recognition · FPHA | Classification accuracy: 93.57 -> 94.09 | Improved |
| Add object label to the action-block input (ablation) | HTT action recognition · H2O | Classification accuracy: 82.65 -> 86.36 | Improved |
| Add hand pose to the action-block input (ablation) | HTT action recognition · FPHA | Classification accuracy: 91.65 -> 94.09 | Improved |
| Add hand pose to the action-block input (ablation) | HTT action recognition · H2O | Classification accuracy: 85.12 -> 86.36 | Improved |
Closest papers by meaning
- On the Utility of 3D Hand Poses for Action Recognition (ECCV 2024 · similarity 0.92)
- First-Person Hand Action Benchmark With RGB-D Videos and 3D Hand Pose Annotations (CVPR 2018 · similarity 0.90)
- Collaborative Learning for 3D Hand-Object Reconstruction and Compositional Action Recognition from Egocentric RGB Videos Using Superquadrics (AAAI 2025 · similarity 0.89)
- ComPose: When to Trust Hands for Object Pose Tracking (NeurIPS 2026 · similarity 0.88)
- Modeling Fine-Grained Hand-Object Dynamics for Egocentric Video Representation Learning (ICLR 2025 · similarity 0.87)
- StableHand: Quality-Aware Flow Matching for World-Space Dual-Hand Motion Estimation from Egocentric Video (NeurIPS 2026 · similarity 0.87)
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.
