Image-Text Matching on EQ-KUBRIC EQBEN
51.9Text ScoreFIBER + EQSIM
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| FIBER + EQSIMEQSIM=true, Fine-tuning=Flickr30K2023.03 | 51.9 | 48.4 | 37.38 | |
| FIBER + FT (F30K)Fine-tuning=Flickr30K2023.03 | 50.31 | 46.06 | 34.66 | |
| FIBERZero-shot=true2023.03 | 47.94 | 45.6 | 33.53 | |
| METERZero-shot=true2023.03 | 44.28 | 35.2 | 27.26 | |
| METER + EQSIMEQSIM=true, Fine-tuning=Flickr30K2023.03 | 43.68 | 39.08 | 28.42 | |
| METER + FT (F30K)Fine-tuning=Flickr30K2023.03 | 41.11 | 34.59 | 24.33 | |
| ViLTZero-shot=true2023.03 | 31.23 | 27 | 17.9 | |
| ALBEFZero-shot=true2023.03 | 29.87 | 30.18 | 18.58 | |
| BLIPZero-shot=true2023.03 | 29.38 | 32.25 | 18.73 | |
| FLAVAZero-shot=true2023.03 | 28.88 | 28.18 | 15.9 | |
| CLIP (RN-50)Zero-shot=true, Backbone=ResNet-502023.03 | 21.33 | 21.93 | 9.75 | |
| CLIP (ViT-B/32)Zero-shot=true, Backbone=ViT-B/322023.03 | 20.56 | 21.29 | 9.66 | |
| LXMERTZero-shot=true2023.03 | 18.5 | 15.35 | 7.26 | |
| VILBERTZero-shot=true2023.03 | 17.55 | 18.44 | 8.13 |