In this chapter, we will implement a facial recognition system using a Siamese neural network. Such facial recognition systems are prevalent in smartphones and other smart security systems in modern buildings and facilities. We will go through the theory behind Siamese neural networks, and why facial recognition problems are a special class of problems in image recognition, making it difficult for a conventional convolutional neural networks (CNNs) to solve them. We will train and implement a robust model that can recognize faces, even when the subject has different expressions and when the photo is taken from different angles. Finally, we will write our own program that uses the pre-trained neural network and a webcam, to authenticate the user sitting in front of the computer.
Specifically, these are the topics that...