Face Encoding

Introduction

Every face embodies plenty of distinguishable landmarks that constitute the facial features. These landmarks, also defined as nodal points, are measured during the face recognition process.

Among the measurements carried out are:

• The distance between the eyes
• The nose width
• The eye socket depth
• The cheekbone shape
• The jawline length

Once measured, a numerical code called face encoding, or a face print, is created to represent the face in question.

Objective

The aim of this lesson is to show the steps needed to generate the faceprints of the faces detected in a digital image while leveraging the capabilities of the new Python library face_recognition.

It’s worth noting that a face encoding generated by the face_recognition Python library is composed of a set of 128 computer-generated measurements representing a particular face. Typically, the generated face encodings for two different faces would vary, whereas the face encodings of different pictures pertaining to the same person would be very close.

The face encoding process is depicted in the following flow diagram:

Dependencies

We’ll be using the following external Python libraries:

Let’s encode the faces!

Let’s look at the main functions of ...