InterHuman
Benchmarks
| Task Name | Dataset Name | SOTA Result | Trend | |
|---|---|---|---|---|
| Interactive Motion Synthesis | InterHuman (test) | R Precision (Top 1)49.1 | 37 | |
| text-conditioned human interaction generation | InterHuman (test) | R Precision (Top 1)49.6 | 27 | |
| Human-human interaction motion generation | InterHuman | FID0.273 | 23 | |
| Text-to-Interaction Motion Generation | InterHuman (test) | Interaction Alignment0.675 | 19 | |
| Human-Human Motion Generation | InterHuman (test) | Top-1 R Precision50.1 | 11 | |
| Human Motion Generation | InterHuman (test) | R@Top368.3 | 10 | |
| Human Action-Reaction Synthesis | InterHuman-AS (test) | RTop372.2 | 9 | |
| Individual Motion Generation | InterHuman | R-Precision56.3 | 7 | |
| Text-to-Motion Generation | InterHuman (test) | R-Precision (Top 1)0.481 | 6 | |
| Human action-reaction synthesis | InterHuman-AS SMPL-X (test) | R Precision (Top 3)0.407 | 6 | |
| Reaction Generation | InterHuman | FID2.055 | 4 | |
| Interactive Two-person Reactive Motion Generation | InterHuman-AS | R-Precision Top 30.629 | 3 | |
| Interactive Two-person Duet Motion Generation | InterHuman-AS | R-Precision Top-145.2 | 3 | |
| Individual Alignment (User Study) | InterHuman (test) | Average Rank1.309 | 3 | |
| Interaction Alignment (User Study) | InterHuman (test) | Average Rank1.182 | 3 | |
| Text-to-motion generation | InterHuman re-annotated single-person text (test) | R-Precision Top 145.2 | 3 | |
| Human Motion Estimation | Interhuman | Joint Error (Local)30.4 | 2 | |
| motion in-betweening editing | InterHuman (test) | R Precision Top 151.6 | 2 | |
| Motion-to-text retrieval | InterHuman (test) | R@18.26 | 2 | |
| Text-to-motion retrieval | InterHuman (test) | R@19.51 | 2 |
Showing 20 of 20 rows