A Survey on Factors Affecting Facial Expression Recognition based on Convolutional Neural Networks

Abstract

Humans are generally good at recognising emotions which are portrayed on another person’s face. Can the same be said for machines? In recent years, there has been a tremendous amount of progress in the field of computer vision using deep learning methods, namely by Convolutional Neural Networks (CNNs). How good are these CNNs at recognising facial expressions? With the explosion of research outputs using CNN for Facial Expression Recognition (FER) in recent years, it is an appropriate time to review the state of the art in this field, provide a critical analysis of what has and has not been achieved, and synthesize recommendations for each step of the process needed for FER. This work serves as a guide to those who are new to the field. This survey provides a critique of past work, highlights recommendations and list some open, unanswered questions in FER that deserve further investigation.