Sentence-to-Image Retrieval on Flickr30k zero-shot generalization from MS-COCO (test)
55.8R@13SHNet
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| 3SHNetVisual Representation=Region+Grid, Ensemble=false2024.04 | 55.8 | 81.4 | 88.5 | 490 | |
| 3SHNetVisual Representation=Region, Ensemble=true2024.04 | 54.5 | 79.5 | 86.8 | 478.6 | |
| DIMEEnsemble=true, Source=derived from released pre-trained model2024.04 | 53.7 | 79.2 | 86.5 | 471.4 | |
| 3SHNetVisual Representation=Grid, Ensemble=false2024.04 | 53.7 | 79.6 | 87 | 477.8 | |
| ESAEnsemble=false2024.04 | 51.5 | 77.9 | 85.7 | 467.4 | |
| VSE∞Ensemble=false, Source=derived from released pre-trained model2024.04 | 50 | 77 | 84.9 | 462.8 | |
| DIMEEnsemble=false, Source=derived from released pre-trained model2024.04 | 49.7 | 76.1 | 83.9 | 453.2 | |
| SGRAFEnsemble=true2024.04 | 48.1 | 73.9 | 81.9 | 450.2 | |
| SGREnsemble=false2024.04 | 40.5 | 68.6 | 77.7 | 404.6 | |
| CVSEEnsemble=false2024.04 | 39.9 | 68.6 | 77.2 | 414.1 |