CAE_ISIC
Convolutional autoencoders (fully convolutional and another with fully connected layer in the middle)
Inside the non-mnist .py files there is a path variable at the beggining that should be changed to match where you put the training images.