Text Retrieval on Flickr8K (test)
84.2R@5Full-IMRAM
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| Full-IMRAMmatching strategy=Full (Image + Text)2020.03 | 84.2 | 54.7 | 91 | 420 | |
| Text-IMRAMmatching strategy=grounded on Text2020.03 | 81.5 | 52.1 | 90.1 | 412.1 | |
| SCANre-run using authors' code=true2020.03 | 81 | 52.2 | 89.2 | 407.4 | |
| Codebook-CLIPPretrain=CC12M, Evaluation Protocol=Zero-shot2024.04 | 81 | — | — | — | |
| Image-IMRAMmatching strategy=grounded on Image2020.03 | 78.1 | 48.5 | 85.3 | 379.2 | |
| IL-CLIPPretrain=CC12M, Evaluation Protocol=Zero-shot2024.04 | 77 | — | — | — | |
| CLIPPretrain=CC12M, Evaluation Protocol=Zero-shot2024.04 | 75 | — | — | — | |
| NegCLIPPretrain=CC12M, Evaluation Protocol=Zero-shot2024.04 | 69 | — | — | — | |
| Codebook-CLIPPretrain=CC12M, Evaluation Protocol=Zero-shot2024.04 | 65 | — | — | — | |
| IL-CLIPPretrain=CC12M, Evaluation Protocol=Zero-shot2024.04 | 63 | — | — | — | |
| CLIPPretrain=CC12M, Evaluation Protocol=Zero-shot2024.04 | 60 | — | — | — | |
| NegCLIPPretrain=CC12M, Evaluation Protocol=Zero-shot2024.04 | 57 | — | — | — | |
| Codebook-CLIPPretrain=CC3M, Evaluation Protocol=Zero-shot2024.04 | 57 | — | — | — | |
| IL-CLIPPretrain=CC3M, Evaluation Protocol=Zero-shot2024.04 | 57 | — | — | — | |
| m-CNN2020.03 | 53.7 | 24.8 | 67.1 | 275.2 | |
| CLIPPretrain=CC3M, Evaluation Protocol=Zero-shot2024.04 | 50 | — | — | — | |
| Codebook-CLIPPretrain=CC3M, Evaluation Protocol=Zero-shot2024.04 | 47 | — | — | — | |
| IL-CLIPPretrain=CC3M, Evaluation Protocol=Zero-shot2024.04 | 46 | — | — | — | |
| NegCLIPPretrain=CC3M, Evaluation Protocol=Zero-shot2024.04 | 42 | — | — | — | |
| CLIPPretrain=CC3M, Evaluation Protocol=Zero-shot2024.04 | 41 | — | — | — | |
| DVSA2020.03 | 40.6 | 16.5 | 54.2 | 199.9 | |
| Codebook-CLIPPretrain=DataComp, Evaluation Protocol=Zero-shot2024.04 | 36 | — | — | — | |
| NegCLIPPretrain=CC3M, Evaluation Protocol=Zero-shot2024.04 | 35 | — | — | — | |
| CLIPPretrain=DataComp, Evaluation Protocol=Zero-shot2024.04 | 31 | — | — | — | |
| IL-CLIPPretrain=DataComp, Evaluation Protocol=Zero-shot2024.04 | 31 | — | — | — | |
| NegCLIPPretrain=DataComp, Evaluation Protocol=Zero-shot2024.04 | 29 | — | — | — | |
| Codebook-CLIPPretrain=DataComp, Evaluation Protocol=Zero-shot2024.04 | 26 | — | — | — | |
| IL-CLIPPretrain=DataComp, Evaluation Protocol=Zero-shot2024.04 | 26 | — | — | — | |
| CLIPPretrain=DataComp, Evaluation Protocol=Zero-shot2024.04 | 24 | — | — | — | |
| NegCLIPPretrain=DataComp, Evaluation Protocol=Zero-shot2024.04 | 23 | — | — | — | |
| DeViSE2020.03 | 16.5 | 4.8 | 27.3 | 104.2 |