cvqluu / Angular-Penalty-Softmax-Losses-Pytorch

Angular penalty loss functions in Pytorch (ArcFace, SphereFace, Additive Margin, CosFace)
MIT License
482 stars 91 forks source link

change the dataset #4

Open iamzhangyunlu opened 4 years ago

iamzhangyunlu commented 4 years ago

Hi there, thank you for your code. Here is a question: I changed the dataset using FER2013 instead of Fashion MNIST, the original fer2013 is csv format, and I changed it into jpg, and the dimension become 3 (original dimension is 1, grey images).

Will that affect the 3D mapping pics? Cuz I achieved bad results by changing the dataset. @cvqluu