bead_cnn.py - training and evaluating the model. Calls bead_img_loader_3classes_multiple_folder and bead_model_478x478_10x. The number of steps is calculated based on the target number of epochs and number of images in the training collection (aim is 10 epochs).

bead_img_loader_3classes_multiple_folders.py - loads the training and validation folders. Calls images contained within folders (droplets containing exactly one cell/object, two cell/objects, or zero cell/objects).

bead_model_478_x_478_10x.py = model function for Convolutional neural network (CNN). Number of CNN layers and kernel sizes are selected here. Inputs dropout rate, estimates accuracy and loss. 

Real_Time_cnn_predictor.py = Calls the trained model and real-time predicts, categorises and sorts the droplets. Triggers camera, predicts and sends electrical pulse for sorting.

test_camera_multiframe_Trig_FG.py - function for image acquisition. Used for acquiring large datasets for training (triggers camera).