Multiple Choice on MSR-VTT
93.5AccuracyInternVideo
Evaluation Results
| Method | Links | |
|---|---|---|
| InternVideoBackbone=ViT, WiSE-FT=true, Zero-shot=true2023.10 | 93.5 | |
| InternVideoBackbone=UniformerV2, WiSE-FT=true, Zero-shot=true2023.10 | 93.2 | |
| InternVideoWiSE-FT=false, Backbone=UniformerV2, Zero-shot=true2023.10 | 92.6 | |
| All-in-oneSupervised Training=true2023.10 | 92.3 | |
| VidLA-B/16mode=zero-shot, backbone=ViT-B/162024.03 | 92.2 | |
| VIOLETSupervised Training=true2023.10 | 91.9 | |
| InternVideomode=zero-shot, backbone=ViT-B2024.03 | 91.3 | |
| MERLOTSupervised Training=true2023.10 | 90.9 | |
| JSFusionSupervised Training=true2023.10 | 83.4 | |
| All-in-onemode=zero-shot2024.03 | 80.3 | |
| All-in-oneSupervised Training=false, Zero-shot=true2023.10 | 80.3 |