Text-to-Image Retrieval on Flickr30k-CN (test)
84.4R@1R2D2
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| R2D2Backbone=ViT-L2022.05 | 84.4 | 96.7 | 98.4 | — | |
| R2D2Backbone=ViT-B2022.05 | 79.2 | 95.2 | 97.3 | — | |
| WukongBackbone=ViT-L2022.05 | 77.4 | 94.5 | 97 | — | |
| CLIPBackbone=ViT-L2022.05 | 77.3 | 94.4 | 97.2 | — | |
| FILIPBackbone=ViT-L2022.05 | 76.9 | 94.9 | 97.4 | — | |
| CLIPBackbone=ViT-B2022.05 | 69 | 90.3 | 95 | — | |
| WukongBackbone=ViT-B2022.05 | 67.6 | 89.6 | 94.2 | — | |
| FILIPBackbone=ViT-B2022.05 | 57.5 | 84.3 | 90.6 | — |