The primary goal of the face recognition network is to recognize a person's identity in a given image. For our task, we will be using the pre-trained Inception-ResNet-2 model without fully connected layers. Keras has a pretty good library of pre-trained models. For experimentation purposes, you can use other networks, such as Inception or ResNet-50, as well. To learn more about Inception-ResNet-2, visit the link https://arxiv.org/pdf/1602.07261.pdf. The pre-trained Inception-ResNet-2 network, once provided with an image, returns the corresponding embedding. The extracted embeddings for the real image and the reconstructed image can be calculated by calculating the Euclidean distance of the embeddings. More on the face recognition network will be covered in the Keras implementation of Age-cGAN section.