[SDD-FIQA] Why does pseudo label value of faces images with mask lower than faces images without mask?

[chenlemax]

Thank you for your greate work!

I get the pseudo label by running gen_pseudo_labels.py. But I found that unmasked faces pseudo label value lower than masked faces. The masked faces images are augmented from the unmasked face. So why does face images with a mask have higher image scores? Is it possible that there are too many face images with the mask of the same person？

[Slinene]

Thanks for your following our work,
It is possible that different person with mask may be more similar than same person w/wo mask. To solve this problem, you can use face recognition with mask or carefully balance the mask and unmask samples during generating pseudo label.