Video Recognition on VidTAB few-shot adaptation
46.9AverageST-Adapter-CLIP-L
Evaluation Results
| Method | Links | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| ST-Adapter-CLIP-LPretrain Data=CLIP2023.10 | 46.9 | 43 | 45 | 31.2 | 39.4 | 49.4 | 64.9 | 72.3 | 29.9 | |
| ZeroI2V-CLIP-LPretrain Data=CLIP2023.10 | 46.5 | 41.3 | 46.8 | 31.2 | 39.3 | 47.2 | 64.6 | 70.6 | 30.6 | |
| CLIP-LPretrain Data=CLIP2023.10 | 42.8 | 31.2 | 38 | 32.3 | 36.3 | 50.3 | 58.5 | 67.7 | 28.1 | |
| ViCLIP-LPretrain Data=CLIP+InternVid200M2023.10 | 42.7 | 36.7 | 43.9 | 30.2 | 36.8 | 46.9 | 54.8 | 65.4 | 27.2 |