Facial emotion recognition using vision transformer

Abstract

The facial emotion recognition task (FER) has gained a lot of attention in the recent years due to the advancement in deep learning and artificial intelligence. The vast number of facial emotion databases made available online also encouraged research and development in this area. Researchers in the field are experimenting various techniques which allow the computer to extract facial features, study them, to build highly accurate prediction models. Analysis and continuous improvements to these image recognition models have produced exceptional results for FER tasks. The aim of this project is to explore a vision transformer deep learning model for the FER task. The proposed model is evaluated on 2 publicly available databases and analysis is done at the different stages of the experiment.