Text-to-Video Retrieval on MSR-VTT 1k-Miech (test)
26.6R@1MMT
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| MMTPre-training=HowTo100M2020.09 | 26.6 | 57.1 | 69.6 | 4 | — | — | |
| Dual EncodingFeatures=Transformer features included2020.09 | 25.7 | 52.9 | 66.4 | 5 | 38.9 | — | |
| Dual EncodingSource=Proposed, Visual Feature=ResNeXt-ResNet2020.09 | 23 | 50.6 | 62.5 | 5 | 36.1 | 277.9 | |
| Dual EncodingFeatures=Standard 2D CNN2020.09 | 23 | 50.6 | 62.5 | 5 | 36.1 | — | |
| HGRSource=Re-trained, Visual Feature=ResNeXt-ResNet2020.09 | 22.9 | 50.2 | 63.6 | 5 | 35.9 | 265.9 | |
| W2VV++Source=Re-trained, Visual Feature=ResNeXt-ResNet2020.09 | 21.7 | 48.6 | 60.9 | 6 | 34.4 | 255.3 | |
| MMTPre-training=None2020.09 | 20.3 | 49.1 | 63.9 | 6 | — | — | |
| CESource=Original Paper2020.09 | 18.2 | 46 | 60.7 | 7 | — | 249.2 | |
| CESource=Re-trained, Visual Feature=ResNeXt-ResNet2020.09 | 17.8 | 42.8 | 56.1 | 8 | 30.3 | 233.1 | |
| TCESource=Original Paper2020.09 | 17.1 | 39.9 | 53.7 | 9 | — | — | |
| VSE++Source=Re-trained, Visual Feature=ResNeXt-ResNet2020.09 | 17 | 40.9 | 52 | 10 | 16.9 | 220.5 | |
| TCESource=Re-trained, Visual Feature=ResNeXt-ResNet2020.09 | 17 | 44.7 | 58.3 | 7 | 30 | 236.6 | |
| MEESource=Original Paper2020.09 | 16.8 | 41 | 54.4 | 9 | — | — | |
| Dual EncodingFeatures=Transformer features only2020.09 | 16.8 | 41.6 | 55.8 | 8 | 28.7 | — | |
| MEESource=Re-trained, Visual Feature=ResNeXt-ResNet2020.09 | 15.7 | 39 | 52.3 | 9 | 27.1 | 218.7 | |
| fpoSESource=Original Paper2020.09 | 14.3 | 38.1 | 53 | 9 | — | 217.5 | |
| W2VVSource=Re-trained, Visual Feature=ResNeXt-ResNet2020.09 | 2.7 | 12.5 | 17.3 | 83 | 7.9 | 145.3 |