Emotion Recognition Through Facial Expressions Using Supervised Learning with Logistic Regression

More than half of a message meaning is conveyed through facial expressions. Different studies have also shown that some of these expressions are considered universal. However, one of the most interesting characteristics of facial expressions is that they show emotion. This characteristic has led several areas to study them with different aims, and computing was no exception. Face detection and the main parts have been one of the major breakthroughs in computer vision. Together with the rise of machine learning in the last decade, it allowed the development of systems capable of detecting emotions through facial expression analysis. This paper shows the stages of development, training and testing of an algorithm based on logistic regression used for emotion detection, including specific details of the optimization process and the results in the training and test set.