This repository contains two Python scripts:

1)trainer.py is the script used to train the weights and yield the reports which are saved in the TrainingReports directory

2)predictor.py loads the model, then loads the weights stored in TrainingReports to predict and generate a class activation map example.

The TrainingReports directory: contains the trained weights for every mueller component and training-validation reports for 3 cross-validation runs.

Pre-requisites: keras numpy pandas sklearn

These scripts were tested and run in Python 3.2 and Python 3.6.