Impact of Deep Learning Approaches on Facial Expression Recognition in Healthcare Industries

A facial expression recognition system that can provide quick assistance to the healthcare system and exceptional services to the patients is proposed in this article. The implementation of this work is divided into three components. In the first component, landmark points on the facial region are detected; a fixed-sized rectangular box is obtained by normalizing the detected face region, and then, down sampled to its varying sizes producing multiresolution images. Different convolution neural network architectures are proposed in the second component for analyzing the textual information within the multiresolution facial images. To extract more discriminating features and enhance the proposed system’s performance, some amalgamation of transfer learning, progressive image resizing, data augmentation, and fine tuning of parameters are employed in the third component. For experimental purposes, three benchmark databases, static facial expressions in the wild, Cohn-Kanade, and Karolinska directed emotional faces, are employed with some existing methods concerning these databases. The comparison with these databases shows the superiority of the proposed system.