Image-to-Text Retrieval on COCO-CN (val)
83.2Recall@1FG-CLIP 2
Evaluation results
| Method | Links | |
|---|---|---|
| FG-CLIP 2Backbone=ViT-So/162025.10 | 83.2 | |
| FG-CLIP 2Backbone=ViT-L/162025.10 | 82.8 | |
| Meta CLIP 2Backbone=ViT-H/142025.10 | 80.1 | |
| FG-CLIP 2Backbone=ViT-B/162025.10 | 77.2 | |
| Chinese-CLIPBackbone=ViT-L/142025.10 | 74.3 | |
| SigLIP 2Backbone=ViT-L/162025.10 | 74.2 | |
| SigLIP 2Backbone=ViT-So/162025.10 | 72 | |
| R2D2Backbone=ViT-L/142025.10 | 69.6 | |
| Chinese-CLIPBackbone=ViT-B/162025.10 | 68.8 | |
| SigLIP 2Backbone=ViT-B/162025.10 | 68.6 | |
| R2D2Backbone=ViT-B/162025.10 | 60.1 |