Testing Siamese net
Chopra network triplet nets
Tests with triplet
Organized the trainer
Triplet selection Data augmentation
New triplet selection
Managing better the GPUs
Fixed problem with the learning rate
Update learning rate issue #7
New load and save methods
Fixed FaceNET
New FaceNET Pooling with strides
A trainer can load a pretrained network
New learning rate
Introduced the VGG16
- Implemented the batch normalization - Removed the hdf5 backend
- Average Pooling - VGG 16 mod
Created the concept of normalizer
GPU configurations