Manta Ray Foraging Optimization with Transfer Learning Driven Facial Emotion Recognition

Abstract

Recently, facial expression-based emotion recognition techniques obtained excellent outcomes in several real-time applications such as healthcare, surveillance, etc. Machine-learning (ML) and deep-learning (DL) approaches can be widely employed for facial image analysis and emotion recognition problems. Therefore, this study develops a Transfer Learning Driven Facial Emotion Recognition for Advanced Driver Assistance System (TLDFER-ADAS) technique. The TLDFER-ADAS technique helps proper driving and determines the different types of drivers’ emotions. The TLDFER-ADAS technique initially performs contrast enhancement procedures to enhance image quality. In the TLDFER-ADAS technique, the Xception model was applied to derive feature vectors. For driver emotion classification, manta ray foraging optimization (MRFO) with the quantum dot neural network (QDNN) model was exploited in this work. The experimental result analysis of the TLDFER-ADAS technique was performed on FER-2013 and CK+ datasets. The comparison study demonstrated the promising performance of the proposed model, with maximum accuracy of 99.31% and 99.29% on FER-2013 and CK+ datasets, respectively.