Text-to-Video Retrieval on Something-to-Something Template v2
90.8R@1UMT-L
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| UMT-L#Pairs=17M2023.03 | 90.8 | 100 | 100 | — | |
| UMT-L#Pairs=25M2023.03 | 90.8 | 99.4 | 100 | — | |
| UMT-L#Pairs=5M2023.03 | 90.2 | 99.4 | 100 | — | |
| UMT-B#Pairs=25M2023.03 | 87.9 | 99.4 | 100 | — | |
| UMT-B#Pairs=5M2023.03 | 87.3 | 100 | 100 | — | |
| UMT-B#Pairs=17M2023.03 | 86.8 | 99.4 | 100 | — | |
| VINDLUPretraining Data=17M2022.12 | 86.2 | 99.4 | — | 84.6 | |
| HiTeA# PT Data=5M2022.12 | 85.6 | 100 | 100 | — | |
| HiTeA#Pairs=5M2023.03 | 85.6 | 100 | 100 | — | |
| VINDLUPretraining Data=25M2022.12 | 83.3 | 100 | — | 84.4 | |
| VINDLU#Pairs=25M2023.03 | 83.3 | 100 | — | — | |
| VINDLUPretraining Data=5M2022.12 | 82.2 | 98.9 | — | 82.7 | |
| SingularityPretraining Data=17M2022.12 | 77.6 | 96 | — | 80 | |
| Singularity#Pairs=17M2023.03 | 77.6 | 96 | — | — | |
| Clip4Clip# PT Data=400M2022.12 | 77 | 96.6 | 98.3 | — | |
| Singularity# PT Data=5M2022.12 | 77 | 98.9 | 99.4 | — | |
| CLIP4ClipPretraining Data=400M2022.12 | 77 | 96.6 | — | 77.9 | |
| CLIP4Clip#Pairs=400M2023.03 | 77 | 96.6 | — | — | |
| Frozen# PT Data=5M2022.12 | 52.9 | 94.8 | 99.4 | — |