Question about the object detection

[Richar-Du]

When encoding the image to prompt, you mentioned captions and bounding boxes, I wonder which object detection model you utilized to generate the bounding boxes?

[wanxinzzz]

When encoding the image to prompt, you mentioned captions and bounding boxes, I wonder which object detection model you utilized to generate the bounding boxes?

I think the bounding boxes come from ground truth in coco dataset

[haotian-liu]

Hi @Richar-Du both annotations come from the original COCO dataset: captions from coco-caption-2014 annotation, and boxes from coco-instances-2014 annotations.

Thanks @wanxinzzz for answering!

[Richar-Du]

Got it, thanks for your explanations :)