H2O
Benchmarks
| Task Name | Dataset Name | SOTA Result | Trend | |
|---|---|---|---|---|
| Action Recognition | H2O (test) | Accuracy91.73 | 26 | |
| 3D Hand Pose Estimation | H2O | MPJPE Both11.4 | 19 | |
| 2D hand trajectory prediction | H2O-PT | ADE0.036 | 9 | |
| Hand Pose Estimation | H2O | Cross-Subject MJE23.8 | 8 | |
| Energy Modeling | H2O (test) | MAE0 | 8 | |
| 3D Hand Pose Estimation | H2O (same-domain) | MPJPE22.77 | 8 | |
| Action Recognition | H2O | Action Accuracy93.39 | 8 | |
| 3D Hand Pose Estimation | H2O (test) | MEPE (Camera Space)35.02 | 8 | |
| Object 6DoF Tracking | H2O | Local RRE23.24 | 7 | |
| Hand Reconstruction | H2O (test) | Jerk2.34 | 7 | |
| Exo2ego Frame Retrieval | H2O | mAP@583.88 | 7 | |
| Ego2exo Frame Retrieval | H2O | mAP@578.21 | 7 | |
| 3D hand motion generation | H2O | Top-1 Accuracy75 | 6 | |
| Interaction Recognition | H2O (test) | Accuracy79.25 | 6 | |
| Interaction Recognition | H2O (val) | Accuracy86.78 | 6 | |
| 2D Hand Pose Estimation | H2O (test) | PCK@0.297.38 | 6 | |
| 3D Hand-Object Pose Estimation | H2O (test) | Object Error (mm)47.9 | 5 | |
| 4D Hand Motion Reconstruction | H2O | G-MPJPE45.6 | 5 | |
| Text-to-Hand-Object Motion Generation | H2O (test) | Top-3 Accuracy99.2 | 5 | |
| Phase Classification | H2O (test) | Accuracy (0.1 tolerance)43.5 | 5 | |
| Novel View Synthesis | H2O subject4_h1 sequences | PSNR18.14 | 4 | |
| Hand Reconstruction | H2O Viewpoint (train) | PSNR24.66 | 4 | |
| Exocentric-to-egocentric translation | H2O (unseen subjects) | FID96.429 | 4 | |
| Exocentric-to-egocentric translation | H2O (unseen scenes) | FID90.893 | 4 | |
| Exocentric-to-egocentric translation | H2O (unseen actions) | FID33.284 | 4 |
Showing 25 of 27 rows