Video-to-Text Retrieval on MSR-VTT 7K/9K (train)
43.1R@1CLIP4Clip-meanP
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| CLIP4Clip-meanPTraining protocol=Training-9K, TrainD=W+M, E2E=false2021.04 | 43.1 | 70.5 | 81.2 | 2 | 12.4 | |
| CLIP4Clip-seqLSTMTraining protocol=Training-9K, TrainD=W+M, E2E=false2021.04 | 42.8 | 71 | 80.4 | 2 | 12.3 | |
| CLIP4Clip-seqTransfTraining protocol=Training-9K, TrainD=W+M, E2E=false2021.04 | 42.7 | 70.9 | 80.6 | 2 | 11.6 | |
| CLIP4Clip-tightTransfTraining protocol=Training-9K, TrainD=W+M, E2E=true2021.04 | 40.6 | 69.5 | 79.5 | 2 | 13.6 | |
| HiTTraining protocol=Training-9K, TrainD=H+M, E2E=false2021.04 | 32.1 | 62.7 | 74.1 | 3 | — | |
| TT-CE+Training protocol=Training-9K, TrainD=M, E2E=false2021.04 | 32.1 | 62.7 | 75 | 3 | — | |
| AVLnetTraining protocol=Training-9K, TrainD=H+M, E2E=false2021.04 | 28.5 | 54.6 | 65.2 | 4 | — | |
| SSBTraining protocol=Training-9K, TrainD=H+M, E2E=false2021.04 | 28.5 | 58.6 | 71.6 | 3 | — | |
| CLIP-straightTraining protocol=Zero-shot, TrainD=W, E2E=true2021.04 | 27.2 | 51.7 | 62.6 | 5 | — | |
| MMTTraining protocol=Training-9K, TrainD=H+M, E2E=false2021.04 | 27 | 57.5 | 69.7 | 3.7 | — | |
| CETraining protocol=Training-9K, TrainD=M, E2E=false2021.04 | 20.6 | 50.3 | 64 | 5.3 | — | |
| HowTo100MTraining protocol=Training-7K, TrainD=H+M, E2E=false2021.04 | 16.8 | 41.7 | 55.1 | 8 | — |