Text-to-Image Retrieval on MSCOCO (R@1, R@5, R@10, R@15)
48.53R@1FAST-GOAL
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| FAST-GOALBackbone=ViT-L/142026.05 | 48.53 | 73.22 | 82.12 | 86.3 | |
| EVA-CLIPBackbone=ViT-L/142026.05 | 48.14 | 71.72 | 80.1 | 84.46 | |
| Long-CLIPBackbone=ViT-L/142026.05 | 46.96 | 71.89 | 80.64 | 85.31 | |
| GOALBackbone=ViT-L/142026.05 | 46.29 | 70.85 | 79.67 | 84.4 | |
| FAST-GOALBackbone=ViT-B/162026.05 | 42.81 | 68.61 | 78.1 | 82.92 | |
| EVA-CLIPBackbone=ViT-B/162026.05 | 42.58 | 67.49 | 76.91 | 81.81 | |
| Long-CLIPBackbone=ViT-B/162026.05 | 40.83 | 66.36 | 76.22 | 81.51 | |
| GOALBackbone=ViT-B/162026.05 | 38.86 | 64.36 | 74.15 | 79.58 | |
| CLIPBackbone=ViT-L/142026.05 | 37.29 | 61.82 | 71.64 | 77.46 | |
| CLIPBackbone=ViT-B/162026.05 | 33.95 | 59.46 | 69.93 | 76 |