Text Retrieval on Flickr30K (1K)
95.4R@1SoftMask++
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| SoftMask++#Img=4M, fine-tuned=true2023.04 | 95.4 | 99.7 | 99.9 | |
| ALIGN#Images=1.2B, Fine-tuning=true2022.02 | 95.3 | 99.8 | 100 | |
| ALIGN#Img=1.2B, fine-tuned=true2023.04 | 95.3 | 99.8 | 100 | |
| CODIS#Img=4M, fine-tuned=true2023.04 | 95.1 | 99.4 | 99.9 | |
| Triple Contrastive Learning (TCL)#Images=4M, Fine-tuning=true2022.02 | 94.9 | 99.5 | 99.8 | |
| TCL#Img=4M, fine-tuned=true2023.04 | 94.9 | 99.5 | 99.8 | |
| ALBEF#Images=4M, Fine-tuning=true2022.02 | 94.3 | 99.4 | 99.8 | |
| ALBEF#Img=4M, fine-tuned=true2023.04 | 94.3 | 99.4 | 99.8 | |
| TCL#Images=4M, Zero-shot=true2022.02 | 93 | 99.1 | 99.6 | |
| CYCLIPEvaluation Protocol=Fine-tuned2022.05 | 92.3 | 97 | 98.4 | |
| CLIPEvaluation Protocol=Fine-tuned2022.05 | 91.9 | 97 | 98 | |
| ALBEF#Images=4M, Zero-shot=true2022.02 | 90.5 | 98.8 | 99.7 | |
| UNIMO#Images=4M, Fine-tuning=true2022.02 | 89.7 | 98.4 | 99.1 | |
| UNIMO#Img=4M, fine-tuned=true2023.04 | 89.7 | 98.4 | 99.1 | |
| ALIGN#Images=1.2B, Zero-shot=true2022.02 | 88.6 | 98.7 | 99.7 | |
| CLIPEvaluation Protocol=Zero-shot2022.05 | 88.2 | 93.9 | 95.8 | |
| CYCLIPEvaluation Protocol=Zero-shot2022.05 | 88.1 | 93.7 | 95.9 | |
| CLIP#Images=400M, Zero-shot=true2022.02 | 88 | 98.7 | 99.4 | |
| VILLA#Images=4M, Fine-tuning=true2022.02 | 87.9 | 97.5 | 98.8 | |
| VILLA#Img=4M, fine-tuned=true2023.04 | 87.9 | 97.5 | 98.8 | |
| UNITER#Images=4M, Fine-tuning=true2022.02 | 87.3 | 98 | 99.2 | |
| UNITER#Img=4M, fine-tuned=true2023.04 | 87.3 | 98 | 99.2 | |
| ImageBERT#Images=6M, Fine-tuning=true2022.02 | 87 | 97.6 | 99.2 | |
| SOHO#Images=200K, Fine-tuning=true2022.02 | 86.5 | 98.1 | 99.3 | |
| SOHO#Img=200K, fine-tuned=true2023.04 | 86.5 | 98.1 | 99.3 | |
| ViLT#Images=4M, Fine-tuning=true2022.02 | 83.5 | 96.7 | 98.6 | |
| ViLT#Img=4M, fine-tuned=true2023.04 | 83.5 | 96.7 | 98.6 | |
| UNITER#Images=4M, Zero-shot=true2022.02 | 80.7 | 95.7 | 98 | |
| ViLT#Images=4M, Zero-shot=true2022.02 | 73.2 | 93.6 | 96.5 | |
| ImageBERT#Images=6M, Zero-shot=true2022.02 | 70.7 | 90.2 | 94 |