Ensemble of Machine Learning Models for an Improved Facial Emotion Recognition

The creation of algorithms that predict emotional recognition is a subject that has been of particular interest by researchers around the world for the last few years, as many computer vision-based systems make use of this information to get an approximation of the emotional state of an individual. This study aims to develop a real-time emotional recognition algorithm based on the facial expression. Our main contributions are the following: This algorithm was tested in a computational tool designed to stimulate the imitation and recognition of emotions of children with Autism Spectrum Disorder based on their facial expressions. By designing an ensemble of machine learning models which separates emotions into different sets, we are able to improve the recognition accuracy. Additionally, the selection of relevant features greatly reduces the execution time of the algorithm, making it feasible for real-time recognition. Testing of different label combinations is yet to be performed in order to further improve the recognition accuracy.