search

Alibaba-MIIL / ASL

Official Pytorch Implementation of: "Asymmetric Loss For Multi-Label Classification"(ICCV, 2021) paper
MIT License
732 stars 102 forks source link

How can I normalize these datasets NUS-WIDE and Open Image? #104

Open ffgg11 opened 2 years ago

ffgg11 commented 2 years ago

A great work. You provide some pretrained models. I want to use these models. What are the means and variances of these datasets?

Did you use the mean and variance of ImageNet ?