Image-to-Text Retrieval on Flickr-CNA (test)
91.5Recall@1FG-CLIP 2
Evaluation Results
| Method | Links | |
|---|---|---|
| FG-CLIP 2Backbone=ViT-So/162025.10 | 91.5 | |
| FG-CLIP 2Backbone=ViT-L/162025.10 | 90.3 | |
| Meta CLIP 2Backbone=ViT-H/142025.10 | 89.3 | |
| FG-CLIP 2Backbone=ViT-B/162025.10 | 85.4 | |
| Chinese-CLIPBackbone=ViT-L/142025.10 | 82.9 | |
| SigLIP 2Backbone=ViT-L/162025.10 | 79.8 | |
| R2D2Backbone=ViT-L/142025.10 | 78.8 | |
| SigLIP 2Backbone=ViT-So/162025.10 | 78.4 | |
| Chinese-CLIPBackbone=ViT-B/162025.10 | 75.8 | |
| SigLIP 2Backbone=ViT-B/162025.10 | 71.7 | |
| R2D2Backbone=ViT-B/162025.10 | 69.7 |