Multiple-choice on MSRVTT (test)
93.4AccuracyInternVideo
Evaluation Results
| Method | Links | |
|---|---|---|
| InternVideoZero-shot=true, Supervised training=false2022.12 | 93.4 | |
| All-in-one-B *Frames=32022.03 | 92.3 | |
| VidLAzero-shot=true, backbone=ViT-B/162024.03 | 92.2 | |
| All-in-one-BFrames=3x32022.03 | 92 | |
| All-in-one-BFrames=32022.03 | 91.4 | |
| InternVideozero-shot=true, backbone=ViT-B2024.03 | 91.3 | |
| ClipBERTFrames=8x22022.03 | 88.2 | |
| ClipBERTZero-shot=true, Supervised training=true2022.12 | 88.2 | |
| ActBERTFrames=322022.03 | 85.7 | |
| ActBERTZero-shot=true, Supervised training=true2022.12 | 85.7 | |
| JSFusionFrames=402022.03 | 83.4 | |
| JSFusionZero-shot=true, Supervised training=false2022.12 | 83.4 | |
| All-in-one-BFrames=3, Zero-shot=true2022.03 | 80.3 | |
| All-in-oneZero-shot=true, Supervised training=false2022.12 | 80.3 | |
| All-in-onezero-shot=true2024.03 | 80.3 |