Towards the automatic assessment of abilities to produce facial expressions: The case study of children with ASD

This work presents a system for the automatic analysis of facial expressions. The system is completely unobtrusive (nothing to wear, no calibration required) and it takes as input video sequences acquired from off-the-shelf devices (e.g., webcam). A computer vision module, made up by three main components aiming at facial landmark detection and tracking, facial action unit extraction and facial expression assessment respectively, processes images acquired by an on-the-shelf camera. The paper demonstrates the system's potential in analysing facial expressions produced by children and it gives also a glimpse about how it can be exploited to assist in the assessment of neurological diseases.