# install the requirements
pip install -r requirements.txt

# downloads the image patches to the data folder
sh get_data.sh

# constructs the dataset i.e. does the train/val splits and mean image creation
python make_dataset.py

# trains the ensembles, 15 of them
sh train_ensembles.sh

# inferences the test data with each model
sh run_ensemble_inference.sh

# performs the blur shift augmentation inferencing
sh run_ensemble_inference_blur.sh

# performs the colour shift augmentation inferening
sh run_ensemble_inference_colour.sh

# subsambles the ensembles for each of the methods using in the paper, Single, MC Dropout and Ensmeble, currently size 10
python subsample_models.py