search

becauseofAI / MobileFace

A face recognition solution on mobile device.
MIT License
746 stars 207 forks source link

What algorithm was used for alignment and cropping #1

Closed simon-rob closed 6 years ago

simon-rob commented 6 years ago

I have only managed to get the MobileFace_Identification_V1 working as the get_face_feature_mxnet.py doesn't work for V2.

I have modified get_face_feature_mxnet.py for V2 inputs (3 channels v 1 ) but i am not getting any sensible results. When comparing for cosine similarities of different faces they are coming as being similar.

I have noticed that the aligned data image pairs are closely cropped and I was wondering if you could share the algorithm/code on how you created the the training data so I can reproduce the cropping before feeding the prediction.

Many Thanks,

Simon

simon-rob commented 6 years ago

I will answer my own question:

You didn't prepare the dataset they are from https://github.com/deepinsight/insightface and the cropping and alignment is done in face_preprocess.py using a similaritity transform.