Let's break down the face recognition problem into smaller steps and subproblems. That way, we can better understand what's going on under the hood of a facial recognition system. A face recognition problem can be broken down into the following smaller subproblems:
- Face detection: Detect and isolate faces in the image. In an image with multiple faces, we need to detect each of them separately. In this step, we should also crop the detected faces from the original input image, to identify them separately.
- Face recognition: For each detected face in the image, we run it through a neural network to classify the subject. Note that we need to repeat this step for each detected face.
Intuitively, this process makes a lot of sense. If we think of how humans recognize faces, we see that it is very similar to the process that we described...