Fig. 4From: Modelling urban networks using Variational AutoencodersStreet network reconstructed (bottom) from the original images (top) using the trained autoencoderBack to article page