Predicting facial expressions using a CNN

Facial expression recognition is a challenging problem because of the variations of faces, lighting, and expressions (mouth, the degree that the eyes are open, and so on) and also the need to develop an architecture and select parameters that can result in consistently high accuracy. This means that the challenge is to not only determine one facial expression correctly in one lighting condition for one person, but to correctly identify all facial expressions for all people with or without glasses, caps, and so on, and in all lighting conditions. The following CNN example categorizes emotion in seven different classifications: Angry, Disgusted, Afraid, Happy, Sad, Surprised, and Neutral. The steps involved in facial expression recognition are as follows:

1. Import functions—Sequential, Conv2D, MaxPooling2D, AvgPooling2D, Dense, Activation...