MobileFaceNet

This is based on my graduation thesis, where I propose the MobileFaceNet, a smaller Convolution Neural Network to perform Facial Recognition. The model was trained based on the technique Distilling the Knowledge in a Neural Network proposed by Geoffrey Hinton, and as a coarse model it was used the pretrained FaceNet from David Sandberg, which achieves over 98% of accuracy on the LFW dataset.

This repository is currently under development