Video Classification on Moments in Time (test)
41.1Top-1 AccuracyVATT-L
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| VATT-L#Param=306.1M, GFLOPs=298002021.08 | 41.1 | 67.7 | — | |
| ViViT-L#Param=88.9M, GFLOPs=14462021.08 | 38 | 64.9 | — | |
| SCT-L#Param=59.89M, GFLOPs=342.62021.08 | 37.3 | 65.1 | — | |
| SCT-M#Param=33.48M, GFLOPs=162.92021.08 | 36.8 | 61.2 | — | |
| I3D#Param=25M2021.08 | 29.5 | 56.1 | — | |
| CoCaEvaluation protocol=Frozen backbone, Spatial features=pre-pool, Resolution=Higher2023.02 | — | — | 47.4 | |
| CoCaEvaluation protocol=Fully finetuned2023.02 | — | — | 49 | |
| ViT-22BEvaluation protocol=Frozen backbone2023.02 | — | — | 44.9 | |
| ViT-eEvaluation protocol=Frozen backbone2023.02 | — | — | 43.6 |