CNN Triplet Loss function implementation for matlab
This is a matlab implementation of CNN (convolutional neural network) triplet loss function, based on the article "FaceNet: A Unified Embedding for Face Recognition and Clustering" Google Inc 2015.
The goal of this code is to use Pre-trained net for classification of new objects, that does not include in the training set, without training the net again. The classification is done based on extractining features from the net, then use nearest neighbor distance to detrmine whether two objects are from the same or different classes. for more information see the article above.
This is only preliminary, and could be improved a lot. I would like to get any suggestions or help for optimize and improving the code. see the code on Git: GitHub: https://github.com/roytalman/TripletLoss.git
Thanks
Roy Talman
roytalman@gmail.com
Copyright (c) Roy Talman 2018