Text-to-Image Retrieval on Flickr30k (R@1, R@5, R@10, R@15)
0.774R@1EVA-CLIP
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| EVA-CLIPBackbone=ViT-L/142026.05 | 0.774 | 0.9398 | 0.9718 | 0.9816 | |
| FAST-GOALBackbone=ViT-L/142026.05 | 0.7672 | 0.9372 | 0.9674 | 0.976 | |
| Long-CLIPBackbone=ViT-L/142026.05 | 0.7622 | 0.9354 | 0.9626 | 0.974 | |
| GOALBackbone=ViT-L/142026.05 | 0.7476 | 0.9266 | 0.9588 | 0.9722 | |
| FineCLIPBackbone=ViT-L/142026.05 | 0.7268 | 0.9156 | 0.9558 | 0.9678 | |
| CLIPBackbone=ViT-L/142026.05 | 0.6538 | 0.8736 | 0.9224 | 0.9468 |