Text-to-Image Retrieval on Flickr30k-CNA (test)
84.9R@1R2D2
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| R2D2Backbone=ViT-L2022.05 | 84.9 | 97 | 98.6 | — | |
| R2D2Backbone=ViT-B2022.05 | 80.5 | 95.6 | 97.7 | — | |
| FG-CLIP 2Backbone=ViT-So/162025.10 | 77.2 | — | — | — | |
| FG-CLIP 2Backbone=ViT-L/162025.10 | 75 | — | — | — | |
| Meta CLIP 2Backbone=ViT-H/142025.10 | 72.2 | — | — | — | |
| FG-CLIP 2Backbone=ViT-B/162025.10 | 69.9 | — | — | — | |
| Chinese-CLIPBackbone=ViT-L/142025.10 | 69.6 | — | — | — | |
| Chinese-CLIPBackbone=ViT-B/162025.10 | 62.4 | — | — | — | |
| R2D2Backbone=ViT-L/142025.10 | 60 | — | — | — | |
| SigLIP 2Backbone=ViT-L/162025.10 | 53.2 | — | — | — | |
| SigLIP 2Backbone=ViT-So/162025.10 | 51.7 | — | — | — | |
| R2D2Backbone=ViT-B/162025.10 | 51.1 | — | — | — | |
| SigLIP 2Backbone=ViT-B/162025.10 | 49.1 | — | — | — |