Verb prediction on VidSitu (test)
44.67Accuracy@1Slow-D
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| Slow-Dbackbone=Slow-D, pretrain=MoDist-K400+VS2022.04 | 44.67 | 74.38 | 18.4 | |
| Slow-Dbackbone=Slow-D, pretrain=MoDist-K4002022.04 | 42.96 | 73.17 | 17.48 | |
| Slow-Dbackbone=Slow-D, pretrain=Sup-K4002022.04 | 38.29 | 69.27 | 18.7 | |
| Slow-Dbackbone=Slow-D, pretrain=CVRL-K400+VS2022.04 | 35.29 | 65.92 | 14.41 | |
| VidSitu-SlowFastBackbone=SF2026.04 | 33.94 | 70.54 | — | |
| Slow-Dbackbone=Slow-D, pretrain=CVRL-K4002022.04 | 32.84 | 61.57 | 13.59 | |
| Slow-Dbackbone=Slow-D2022.04 | 31.69 | 68.64 | 5.68 | |
| Slowbackbone=Slow, pretrain=Sup-K4002022.04 | 29.05 | 58.69 | 19.19 |