Keystep recognition on COIN
57.2AccuracyOurs
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| OursDownstream architecture=Transformer2023.07 | 57.2 | — | |
| DistantSup.Downstream architecture=Transformer2023.07 | 57 | — | |
| OursDownstream architecture=MLP2023.07 | 55.2 | — | |
| TimeSformerDownstream architecture=Transformer2023.07 | 54.6 | — | |
| DistantSup.Downstream architecture=MLP2023.07 | 54.1 | — | |
| VideoCLIPDownstream architecture=Transformer2023.07 | 51.2 | — | |
| TimeSformerDownstream architecture=MLP2023.07 | 48.3 | — | |
| SlowFastDownstream architecture=Transformer2023.07 | 39.6 | — | |
| VideoCLIPDownstream architecture=MLP2023.07 | 39.4 | — | |
| S3DDownstream architecture=MLP2023.07 | 37.5 | — | |
| S3DDownstream architecture=Transformer2023.07 | 37.3 | — | |
| TSNDownstream architecture=MLP2023.07 | 36.5 | — | |
| SlowFastDownstream architecture=MLP2023.07 | 32.9 | — | |
| ClipBERTDownstream architecture=MLP2023.07 | 30.8 | — |