Psychological Stress Detection by 2D and 3D Facial Image Processing

This work aims to identify people psychological stress through the capture of micro modifications and motions within their facial expression. Exogenous and endogenous causes of stress, from environment and/or psychological conditions that could induce stress, have been reproduced in the experimental test involving real subjects, and their face expressions have been recorded by 2D and 3D image capturing tools to create a sample of emotional database. Successively, 2D and 3D analyses have been performed on recorded data according to the respective protocols, by deep learning and machine learning techniques, and a data driven model of the databases has been developed by neural network approach, to classify the psychobehavioral answers to the different kinds of stress conditions induced on tested people. The ultimate aim of the study is to demonstrate the possibility to analyze data collected on participants from 2D shooting and 3D scans in a consistent way by means of deep learning and machine learning techniques, so that to provide a methodology to identify and classify some of the subtle facial micro-expressions of people involved in stressing activities.