Image-Text Retrieval on Flickr30K 1K (test)
85.6IR@1ALBEF (14M)
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| ALBEF (14M)Pre-training dataset size=> 10M images, Fine-tuning=true2022.10 | 85.6 | 97.5 | 98.9 | 95.9 | 99.8 | 100 | |
| TCLPre-training dataset size=< 10M images, Model scale=Base, Fine-tuning=true2022.10 | 84 | 96.7 | 98.5 | 94.9 | 99.5 | 99.8 | |
| MAPPre-training dataset size=< 10M images, Model scale=Base, Fine-tuning=true2022.10 | 83.8 | 97.2 | 98.7 | 94.9 | 99.5 | 99.8 | |
| ALBEF (4M)Pre-training dataset size=< 10M images, Model scale=Base, Fine-tuning=true2022.10 | 82.8 | 96.7 | 98.4 | 94.3 | 99.4 | 99.8 | |
| METERPre-training dataset size=< 10M images, Model scale=Base, Fine-tuning=true2022.10 | 82.2 | 96.3 | 98.4 | 94.3 | 99.6 | 99.9 | |
| UNIMO-LargePre-training dataset size=< 10M images, Model scale=Large, Fine-tuning=true2022.10 | 78 | 94.2 | 97.1 | 89.4 | 98.9 | 99.8 | |
| VILLA-LargePre-training dataset size=< 10M images, Model scale=Large, Fine-tuning=true2022.10 | 76.3 | 94.2 | 96.8 | 87.9 | 97.5 | 98.8 | |
| UNITER-LargePre-training dataset size=< 10M images, Model scale=Large, Fine-tuning=true2022.10 | 75.6 | 94.1 | 96.8 | 87.3 | 98 | 99.2 | |
| UNITER-BasePre-training dataset size=< 10M images, Model scale=Base, Fine-tuning=true2022.10 | 72.5 | 92.4 | 96.8 | 85.9 | 97.1 | 98.8 | |
| ViLTPre-training dataset size=< 10M images, Model scale=Base, Fine-tuning=true2022.10 | 64.4 | 88.7 | 93.8 | 83.5 | 96.7 | 98.6 |