Machine Learning based Solution for Predicting the Affective State of Children with Autism

Autism, or autism spectrum disorder (ASD), is a development disorder characterized by difficulties in communication and speech and it cannot be cured entirely. The hardest part for individuals with autism is to express their feelings and emotions towards others. This paper describes a mobile application developed using machine learning algorithms for multiclass image classification that needs as input just children's drawings. The solution is dedicated specially to children between 2 and 5 years old and predicts their emotional state. The application has been proved to be robust, providing a 80.6% accuracy of emotional identification, using Firebase AutoML.