Information regarding training/valid performance

[yangfanliang]

Hi! I hope you are doing well.

I have been searching for face attribute classifier and your one is the best I have found. I just wonder if it is possible for you to share some information regarding the dateset you used for training/validation and their corresponding performance (accuracy)?

Any related information would be much appreciated!

[dchen236]

Please read the referenced paper from GitHub, https://arxiv.org/abs/1908.04913 <https://arxiv.org/abs/1908.04913> Regards, Danni Chen UCLA SAMUELI SCHOOL OF ENGINEERING Los Angeles, CA 90095-1600 www.samueli.ucla.edu <http://www.samueli.ucla.edu/>

…

On Jul 17, 2020, at 7:29 AM, sidneyliang ***@***.***> wrote: Hi! I hope you are doing well. I have been searching for face attribute classifier and your one is the best I have found. I just wonder if it is possible for you to share some information regarding the dateset you used for training/validation and their corresponding performance (accuracy)? Any related information would be much appreciated! — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#2>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AH3SGPB7FM6SUJG4NCBMIHLR4BN4TANCNFSM4O6LD5YQ>.

[yangfanliang]

Got it, that is very helpful! I did not read the paper thoroughly enough.

[djalusic]

Hello @dchen236, have you trained the model from scratch on FairFace or did you use transfer learning?