Owl-vit inference quality metrics

[gfatigati]

I’m successully running an inference on a set of images by using Owl-Vit. I get a lot of Confidence prediction among images.

Question: is there a way to evaluate general performance of prediction like AP metric or something else?

Thanks.

[sreerag-datcarts]

I was planning to do the same. @gfatigati Were you able to do this? Are there any resources I can look at?

[gfatigati]

Not yet