Surgical Robots in Smart Hospitals

22A Deep Learning–Based Expression Recognition Analysis Assistant for Psychologists

Harjyot Singh Bagga, Sanjoi Sethi, Rishab Goswami, Annapurna Jonnalagadda^* and Ushus E.Z.

School of Computer Science and Engineering, Vellore Institute of Technology, Vellore, Tamil Nadu, India

Abstract

The stringent use of digital technologies by people fostered the need for a therapist or counsellor who can understand/address troublesome behaviors, emotions, interpersonal concerns and/or physiological reactions. The counsellor conducts various counselling sessions to empathize the problem of the patient and provides a reasonable solution. These counselling sessions mainly happens in three modes: oral, writing and expressions. This research aims to decipher the human’s inner sentiments using three modes of information and without being swayed by the biased opinions of the therapist. In order to achieve this technological goal, we deploy Deep Learning powered algorithms like text-based encoder–decoder transformers, modified vision-based residual networks and time-distributed CNNs to extract the underlying information. It considers humongous data backing and pattern recognition to help give an unbiased accurate representation of the users’ emotions. To build and ship this as a market solution, we build an end-to-end web-based solution to make it ready to use by the psychologist and as a self.

Keywords: Machine learning, computer vision, natural language processing, clinical psychology

22.1 ...