Image-to-Text Retrieval on AIC-ICC (test)
65.4R@1R2D2
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| R2D2Backbone=ViT-L2022.05 | 65.4 | 80.3 | 84.7 | 74.8 | |
| WukongBackbone=ViT-L2022.05 | 61.6 | 80.5 | 86.1 | 71.6 | |
| CLIPBackbone=ViT-L2022.05 | 59.1 | 79.5 | 85.2 | 69.9 | |
| R2D2Backbone=ViT-B2022.05 | 56.8 | 76.2 | 82.1 | 69.3 | |
| FILIPBackbone=ViT-L2022.05 | 54.1 | 75.8 | 82.8 | 67.4 | |
| CLIPBackbone=ViT-B2022.05 | 50.5 | 73 | 80.2 | 63.1 | |
| WukongBackbone=ViT-B2022.05 | 47.5 | 70.6 | 78.6 | 57 | |
| BriVL2022.05 | 45.6 | 68 | 76.3 | 58.7 | |
| FILIPBackbone=ViT-B2022.05 | 42.5 | 67.2 | 76 | 57.6 |