search

jshilong / GPT4RoI

GPT4RoI: Instruction Tuning Large Language Model on Region-of-Interest
Other
506 stars 25 forks source link

Question in Flickr30k dataset #22

Closed hanoonaR closed 1 year ago

hanoonaR commented 1 year ago

Hi Authors,

I was working with the flickr30k dataset and noticed that it returns the original bounding boxes (ori_bboxes) directly, whereas other referring expression datasets utilize selected bounding boxes (select_bboxes) to replace ori_boxes.

ori_bboxes = torch.cat([ori_bboxes], dim=0)

Wouldn't this result in a mismatch between the region questions and the bounding boxes? Could you shed some light on this? Am I missing something?

Thank you!

jshilong commented 1 year ago

Thank you very much for pointing out this issue.

I have checked it and it appears to be a problem caused by my refactoring before the release. I have fixed it. Could you please take a look at the current code to confirm if it is correct? Thank you again.