Few-shot Action Recognition on NTU 120
32.2Accuracy (20 classes)uDTW
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| uDTWLearning Setting=Supervised, Evaluation Protocol=One-shot2026.05 | 32.2 | 39 | 41.2 | 45.3 | 49 | |
| TAPLearning Setting=Supervised, Evaluation Protocol=One-shot2026.05 | 31.2 | 37.7 | 40.9 | 44.5 | 47.3 | |
| sDTW div.Learning Setting=Supervised, Evaluation Protocol=One-shot2026.05 | 30.8 | 38.1 | 40 | 44.7 | 47.3 | |
| sDTWLearning Setting=Supervised, Evaluation Protocol=One-shot2026.05 | 30.3 | 37.2 | 39.7 | 44 | 46.8 | |
| uDTWLearning Setting=Unsupervised, Evaluation Protocol=One-shot2026.05 | 22.7 | 28.3 | 35.9 | 39.4 | 44 | |
| ProtoNetLearning Setting=Supervised, Evaluation Protocol=One-shot2026.05 | 21.7 | 24 | 25.9 | 29.2 | 32.1 | |
| sDTW div.Learning Setting=Unsupervised, Evaluation Protocol=One-shot2026.05 | 20.8 | 26 | 33.2 | 37.5 | 42.3 | |
| MatchNetsLearning Setting=Supervised, Evaluation Protocol=One-shot2026.05 | 20.5 | 23.4 | 25.1 | 28.7 | 30 | |
| sDTWLearning Setting=Unsupervised, Evaluation Protocol=One-shot2026.05 | 20.1 | 25.3 | 32 | 36.9 | 40.9 | |
| EuclideanLearning Setting=Supervised, Evaluation Protocol=One-shot2026.05 | 18.7 | 21.3 | 24.9 | 27.5 | 30 | |
| EuclideanLearning Setting=Unsupervised, Evaluation Protocol=One-shot2026.05 | 13.5 | 16.3 | 20 | 24.9 | 26.2 |