Deep Learning Techniques for Recognizing Facial Expressions on Masked Faces

Abstract

With the global spread of COVID-19, wearing masks has become a routine part of daily life, presenting unique challenges for facial expression recognition technology. Masks cover significant portions of the face, notably reducing the accuracy of traditional facial expression recognition techniques and particularly hampering the interpretation of non-verbal communication. Current facial expression recognition technologies primarily rely on the features of the entire face, but the presence of masks obscures critical areas such as the mouth and cheeks, diminishing the effectiveness of these systems. Consequently, there is a pressing need for a detailed analysis of the visible parts of the face, such as the subtle movements of the eyes and eyebrows, as well as the texture of the skin that remains uncovered by masks. This research aims to address these challenges by developing new technologies capable of effectively recognizing facial expressions even when masks are worn. The study focuses on learning these fine features and seeks to contribute to the construction of advanced facial expression recognition models. Ultimately, this research endeavors to enhance the accuracy and reliability of facial expression recognition technologies in a world where mask-wearing has become commonplace, thereby improving our understanding and interpretation of non-verbal communication.