Image-to-Text Retrieval on COCO 1K (test)
87.4R@1VACSR
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| VACSRBackbone=CLIP ViT-B/162026.05 | 87.4 | 98.2 | 99.4 | |
| PCME++Backbone=CLIP ViT-B/162026.05 | 85.3 | 97.9 | 99.3 | |
| VACSRBackbone=CLIP ViT-B/322026.05 | 84.2 | 97.2 | 99 | |
| PCMEBackbone=CLIP ViT-B/162026.05 | 83.6 | 97.7 | 99.3 | |
| PCME++Backbone=CLIP ViT-B/322026.05 | 81.6 | 97 | 99 | |
| PCMEBackbone=CLIP ViT-B/322026.05 | 80.1 | 96.6 | 98.7 | |
| DAABackbone=CLIP ViT-B/322026.05 | 79.8 | 96.5 | 98.9 | |
| P2RMBackbone=CLIP ViT-B/322026.05 | 78.6 | 96.1 | 98.6 | |
| P2RMBackbone=CLIP ViT-B/162026.05 | 78.3 | 96.2 | 98.7 | |
| DAABackbone=CLIP ViT-B/162026.05 | 46.1 | 76.8 | 87.6 |