Análise facial para detecção de estresse e tédio em jogos digitais

Abstract

The classification of facial expressions using computer vision, in order to obtain a remote extraction of the individual’s physiological signals, is usually related to methods that stimulate the user to have an emotional involvement or to sketch some physiological reaction, such as using, for example, videos and images. Generally, the models used are built to identify a large part of the emotional behaviors of a given group, which causes the behavioral individualities of each person to be disregarded. In the course of this work, we present a method where it is possible to analyze and classify the emotional states of each individual remotely, training a neural network that generates an exclusive model for each of the users, through calibration games. With this, it will be possible to analyze and classify the emotional state of a player, maintaining their behavioral peculiarities.