Facial Micro Expression Detection Using Deep Learning Architecture

Micro expressions are facial expressions that result as a response to repressed emotions in an involuntary manner that lasts for one-fifth of a second. The reading and interpretation of these expressions is of paramount importance to understand nonverbal human behaviors. Using a deep learning architecture that extracts the required features from the FERC-2013 database, six different micro expressions have been recognized and the accuracy of this model has been tested on test images of varied ethnicities and age groups. The paper also successfully experiments the emotion detection process on both individual and group pictures.