hi @anson0910 ,
I noticed that the positive samples are generated by aflw.py. Right? However this script only output positive_16 ~ positive_17, rather than 1-15.
So I wonder what script positive 1-15 are generated from?
Hi hiyijian,
I used datasets such as CACD, feret, 10kUS, and Caltech before.
However, according to my experiments, using the AFLW dataset can achieve adequate accuracy.
hi @anson0910 , I noticed that the positive samples are generated by aflw.py. Right? However this script only output positive_16 ~ positive_17, rather than 1-15. So I wonder what script positive 1-15 are generated from?