Two step convolutional neural network
Steps to follow:
1) Run the Init_folders.py to create the folders
2) Download the vgg weights file to ./weights folder
link - https://github.com/fchollet/deep-learning-models/releases/download/v0.1/vgg16_weights_tf_dim_ordering_tf_kernels.h5
3) Copy the images of of training, validation and test to ./Data/NO/CNN1/imagetrain,./Data/NO/CNN1/imageval,./Data/NO/CNN1/imagetest respectively
4) Copy corresponding annotations to ./Data/NO/CNN1/annotrain,./Data/NO/CNN1/annoval,./Data/NO/CNN1/annotest
annotations are in the form of .png files, with glottis region pixels labelled as '1' and non glottal regions as '0'
5) Train CNN1 - run train1.py with mname=CNN1
6) Get CNN1 output - run predict.py with mname=CNN1
7) Run Boundingbox.m to generate data for CNN2
8) Train CNN2 - run train1.py with mname=CNN2
9) Test the two step CNN method - Run test.py to get the dice scores