Deep Learning for Computer Vision

Computing the similarity between faces

The computing of face similarities is a multi-step problem. The faces have to be detected, followed by finding the fiducial points. The faces can be aligned with the fiducial points. The aligned face can be used for comparison. As I have mentioned earlier, face detection is similar to object detection. So, in order to find the similarities between faces, we will first import the required libraries and also the facenet library, with the help of the following code:

from scipy import miscimport tensorflow as tfimport numpy as npimport osimport facenetprint facenetfrom facenet import load_model, prewhitenimport align.detect_face

The images can be loaded and aligned as shown:

def load_and_align_data(image_paths ...