Can anyone reproduce results similar to those claimed in the paper? #4
Description
I used the Adaface repo and use the backbone GhostFaceNetsV2 from this repo for training, but my results are like 10% worse compared to the paper
{{ message }}
I used the Adaface repo and use the backbone GhostFaceNetsV2 from this repo for training, but my results are like 10% worse compared to the paper