search

clcarwin / sphereface_pytorch

A PyTorch Implementation of SphereFace.
MIT License
714 stars 172 forks source link

Accuracy problem #31

Closed wyq0227 closed 6 years ago

wyq0227 commented 6 years ago

I used my own train.py to train the model which has same data processing and same parameters. But my test result is 98.38% using clcarwin's lfw_eval.py in Python2.7 and Pytorch0.4.0. I don't know why my result is lower than clcarwin's. Does anybody can solve my problem?Thanks.

Gerkam commented 6 years ago

@wyq0227 How many epochs and iterations have you done?

wyq0227 commented 6 years ago

@Gerkam 20, the same as clcarwin's. I found that the problem was that I didn't change the images' channels. I read them in RGB in training process but I read them in BGR in testing process. So there is a little deviation in accuracy.