Action Localization on BABEL Subset-2 v1.0 (test)
65.2mAP@0.1RVTCLR + Skeleton-Snippet Contrastive Learning
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| RVTCLR + Skeleton-Snippet Contrastive LearningEvaluation Protocol=Linear evaluation2025.12 | 65.2 | 57.8 | 49 | 40.4 | 32.9 | 49.1 | |
| AimCLR + Skeleton-Snippet Contrastive LearningEvaluation Protocol=Linear evaluation2025.12 | 64.2 | 55.4 | 46.4 | 38.6 | 28.3 | 46.6 | |
| HiCLR + Skeleton-Snippet Contrastive LearningEvaluation Protocol=Linear evaluation2025.12 | 63.7 | 57.1 | 48.5 | 38.5 | 29.2 | 47.4 | |
| AimCLREvaluation Protocol=Linear evaluation2025.12 | 62.1 | 53.2 | 41.9 | 33.8 | 25.2 | 43.3 | |
| HiCLREvaluation Protocol=Linear evaluation2025.12 | 61.1 | 50.9 | 40.7 | 30.4 | 22.2 | 41.1 | |
| RVTCLR + OursEvaluation Protocol=KNN evaluation, Feature Enhancement=Skeleton-Snippet Contrastive Learning2025.12 | 59.6 | 55 | 46.2 | 40.3 | 30.1 | 46.2 | |
| RVTCLREvaluation Protocol=Linear evaluation2025.12 | 57.7 | 48.5 | 38.2 | 27.9 | 20.7 | 38.8 | |
| HiCLR + OursEvaluation Protocol=KNN evaluation, Feature Enhancement=Skeleton-Snippet Contrastive Learning2025.12 | 56.7 | 51.2 | 43.8 | 35.5 | 29.6 | 43.4 | |
| HiCLREvaluation Protocol=KNN evaluation2025.12 | 55 | 49.4 | 42 | 33 | 25.4 | 41 | |
| AimCLR + OursEvaluation Protocol=KNN evaluation, Feature Enhancement=Skeleton-Snippet Contrastive Learning2025.12 | 54.6 | 50 | 43.7 | 35.9 | 28.7 | 42.6 | |
| RVTCLREvaluation Protocol=KNN evaluation2025.12 | 52.3 | 44.2 | 35.7 | 27.4 | 21.8 | 36.3 | |
| AimCLREvaluation Protocol=KNN evaluation2025.12 | 48.8 | 43.6 | 38 | 31.6 | 23.4 | 37.1 |