Image-to-text retrieval on Sydney
21.3R@1ToMA-domain
Evaluation results
| Method | Links | |
|---|---|---|
| ToMA-domainData=L=U2026.04 | 21.3 | |
| CLIP (fine-tuned)Data=L2026.04 | 20.7 | |
| Soft-PLData=L=U2026.04 | 20.7 | |
| ToMAData=L=U2026.04 | 20.7 | |
| S-CLIPData=L̸=U2026.04 | 20.4 | |
| SemiCLIPData=L̸=U2026.04 | 20.1 | |
| ToMA-domainData=L̸=U2026.04 | 19.5 | |
| Hard-PLData=L̸=U2026.04 | 19 | |
| SemiCLIPData=L=U2026.04 | 18.4 | |
| ToMAData=L̸=U2026.04 | 18.4 | |
| Hard-PLData=L=U2026.04 | 17.9 | |
| S-CLIPData=L=U2026.04 | 17.8 | |
| Soft-PLData=L̸=U2026.04 | 16.7 | |
| S-CLIPData=L=U2023.05 | 14.9 | |
| Hard-PLData=L=U2023.05 | 14.4 | |
| S-CLIPData=L≠U2023.05 | 13.8 | |
| Hard-PLData=L≠U2023.05 | 13.4 | |
| CLIP (fine-tune)Data=L2023.05 | 13.2 | |
| Soft-PLData=L=U2023.05 | 11.5 | |
| Soft-PLData=L≠U2023.05 | 10.9 | |
| CLIP (original)2023.05 | 10.3 | |
| CLIP (original)Data=-2026.04 | 10.3 |