A Robust-Facial Expressions Recognition System using Deep Learning Architectures

In recent times, Facial Expression Recognition strategies are viewed as the main field of examination in biometric innovation. In this exploration paper, we present a Facial Expression Recognition (FER) framework partitioned into three stages: The Viola-Jones face detection algorithm, facial image enhancement utilizing Adaptive Histogram Equalization calculation (AHE), and component learning for classification. For learning the highlights followed by classification we utilized VGG16, ResNet50 Convolutional Neural Networks (CNN) models for the proposed framework. Our trial work was performed on the JAFEE data set and CK+ information base. At last, the correlation with different strategies on the two data sets shows the heartiness and adequacy of the proposed approach. The ResNet50 architecture achieved high accuracy rates of 98.21% and 97.08^% for the respective evaluations.