This is the delivery of an inidividual project as evaluation of the Deep Learning course. The aim of the project is to separate an image obtained as a sum of two images into its components. The two images summed together come from different datasets: mnist and fashion_mnist, respectively.
To do so, a neural network was implemented. The network takes in input the sum of the images and returns the predicted components.
The metric used to evaluate the project is the mean squared error between predicted and ground truth images.