search

salesforce / ALBEF

Code for ALBEF: a new vision-language pre-training method
BSD 3-Clause "New" or "Revised" License
1.57k stars 199 forks source link

grounding results #59

Closed ziyanyang closed 2 years ago

ziyanyang commented 2 years ago

Hi,

I have checked the coco.json file used for pretraining, and I found the images in refcoco+_val.json and refcoco+_test.json files also appear in this pretraining file. Does it mean these images are already "seen" during pretraining step?

LiJunnan1992 commented 2 years ago

Yes, we have discussed this in Appendix A - Visual Grounding. Thanks!

ziyanyang commented 2 years ago

Yes, we have discussed this in Appendix A - Visual Grounding. Thanks!

Got it. Thanks!