search

deepinsight / insightface

State-of-the-art 2D and 3D Face Analysis Project
https://insightface.ai
23.16k stars 5.39k forks source link

extra loss #25

Closed zhly0 closed 6 years ago

zhly0 commented 6 years ago

Thanks for your work,but the extra loss I do not know why you do this: 1 The extra loss is makes the l2 norm small,and you do this for hard sample mining? 2 In the code,the extra loss is used only when the loss is softmax or sphereface,why?The other loss can not use it?

nttstar commented 6 years ago

extra loss is used for joint training, not limited to feature incay.

zhly0 commented 6 years ago

@nttstar thanks for reply! what do you mean by joint training?you mean training different dataset using different fc?