Similar to other GANs, pix2pix is made up of two networks: a generator and a discriminator network. The architecture of the generator network is inspired by the architecture of U-Net (https://arxiv.org/pdf/1505.04597.pdf). Similarly, the architecture of the discriminator network is inspired by the architecture of PatchGAN (https://arxiv.org/pdf/1604.04382.pdf). Both networks are deep convolutional neural networks. In this section, we will be exploring pix2pix in detail.