LeeBC2298

I think the implementation of SphereFace is wrong, because in the original paper of SphereFace: 1. The hyperparameter 'm', which means the angular restrain, should be no less than 3...

About the title

1

I think this project is about 'SphereFace', not 'CosFace'. CosFace use LMCL loss function, not A-softmax loss.

About the range of cosine

1

I think it is better to limit the range of cosine like: cosine = F.linear(F.normalize(input), F.normalize(self.weight)).clamp(-1+eps,1-eps), because in my experiment, when cosine(theta) == 1 the loss would become NaN.

detect.py里好像没考虑宽高哪个长的问题？输入图片是（240*320）的话resize比例应该还是1吧，那怎么输入（320*240）的网络呢

How to control appearance and shape?

2

Hello, I am a freshman in GAN and I am confused with some operations. 1. How can I know which appearance code is for which color? For example, if I...

Recently I used the pytorch version for some simple backdoor attacks, and it really works well. However, I have a question when reading your paper and the code: In sec...