Facial expression based emotion recognition using neural networks

Official URL: http://dx.doi.org/10.1007/978-3-319-93000-8_24

Facial emotion recognition has been extensively studied over the last decade due to its various applications in the fields such as human-computer interaction and data analytics. In this paper, we develop a facial emotion recognition approach to classify seven emotional states (joy, sadness, surprise, anger, fear, disgust and neutral). Seventeen action units tracked by Kinect v2 sensor have been used as features. Classification of emotions was performed by artificial neural networks (ANNs). Six subjects took part in the experiment. We have achieved average accuracy of 95.8% for the case in which we tested our approach with the same volunteers took part in our data generation process. We also evaluated the performance of the network with additional volunteers who were not part of the training data and achieved 67.03% classification accuracy.