Text-to-Image Retrieval on MUGE (test)
60.1R@1R2D2
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| R2D2Backbone=ViT-L2022.05 | 60.1 | 82.9 | 89.4 | 77.5 | |
| R2D2Backbone=ViT-B2022.05 | 53.4 | 78.1 | 86 | 72.5 | |
| WukongBackbone=ViT-L2022.05 | 52.7 | 77.9 | 85.6 | 72.1 | |
| CLIPBackbone=ViT-L2022.05 | 50.1 | 76.9 | 84.9 | 70.6 | |
| CLIPBackbone=ViT-B2022.05 | 43.5 | 71.7 | 80.6 | 65.3 | |
| FILIPBackbone=ViT-L2022.05 | 43.5 | 71.5 | 80.9 | 65.3 | |
| WukongBackbone=ViT-B2022.05 | 39.2 | 66.9 | 77.4 | 61.2 | |
| FILIPBackbone=ViT-B2022.05 | 30.6 | 58.2 | 70.2 | 53 |