Text-to-Image Retrieval on COCO-CN (val)
68.1Recall@1FG-CLIP 2
Evaluation Results
| Method | Links | |
|---|---|---|
| FG-CLIP 2Backbone=ViT-So/162025.10 | 68.1 | |
| FG-CLIP 2Backbone=ViT-L/162025.10 | 66.5 | |
| Meta CLIP 2Backbone=ViT-H/142025.10 | 63.1 | |
| FG-CLIP 2Backbone=ViT-B/162025.10 | 62.9 | |
| Chinese-CLIPBackbone=ViT-L/142025.10 | 59.9 | |
| Chinese-CLIPBackbone=ViT-B/162025.10 | 54.9 | |
| R2D2Backbone=ViT-L/142025.10 | 52.7 | |
| SigLIP 2Backbone=ViT-L/162025.10 | 51.7 | |
| SigLIP 2Backbone=ViT-So/162025.10 | 50.7 | |
| SigLIP 2Backbone=ViT-B/162025.10 | 46.2 | |
| R2D2Backbone=ViT-B/162025.10 | 45.5 |