Fig. 3From: Retinal image synthesis from multiple-landmarks input with generative adversarial networksDiscriminator networkBack to article page