Overfitting consists of memorizing the images of the training set, with the consequent stiffness of the model to make predictions in new images