Image-to-Text Retrieval on DOCCI-CN
79.7Recall@1FG-CLIP 2
Evaluation results
| Method | Links | |
|---|---|---|
| FG-CLIP 2Backbone=ViT-So/162025.10 | 79.7 | |
| FG-CLIP 2Backbone=ViT-L/162025.10 | 77.6 | |
| Meta CLIP 2Backbone=ViT-H/142025.10 | 73.8 | |
| FG-CLIP 2Backbone=ViT-B/162025.10 | 71.2 | |
| Chinese-CLIPBackbone=ViT-L/142025.10 | 49.7 | |
| R2D2Backbone=ViT-L/142025.10 | 49.5 | |
| Chinese-CLIPBackbone=ViT-B/162025.10 | 44.6 | |
| R2D2Backbone=ViT-B/162025.10 | 36.1 | |
| SigLIP 2Backbone=ViT-So/162025.10 | 25 | |
| SigLIP 2Backbone=ViT-L/162025.10 | 24.6 | |
| SigLIP 2Backbone=ViT-B/162025.10 | 7.6 |