Practical Example: Face Recognition

Face recognition is a computer vision problem to detect and identify human faces in an image or video. The first step of facial recognition is to detect and locate the position of the face in the input image. This is a typical object detection task like we learned about in the previous chapters. After the face is detected, a feature set, also called a facial footprint or face embedding, is created from various key points on the face. A human face has 80 nodal points or distinguishing landmarks that are used to create the feature set (USPTO Patent Number US7634662B2, https://patents.google.com/patent/US7634662B2/). The face embedding is then compared against a database to establish the identity of the face.