In this branch I removed the option to give a specific size for the image, and instead limited the choices to 'crop', 'reshape' or 'padding'. As such, the user can choose in which way the images should be changed in order to fit the image size to the depth of the UNet.
As discussed previously, giving the option to modify the size of the images in the dataset should potentially be a process made elsewhere, or at least seperately from the Augmentation and Dataset classes/functions.
The aim of the Augmentation and Dataset is to 'make the code run' by adapting the images to the UNet.