Facial Expression Recognition with Combination of Geometric and Textural Domain Features Extractor using CNN and Machine Learning

Abstract

Human emotions may be characterized by the expressions on their faces, which is the goal of research in FER (facial expression recognition). These include biometric security and human-computer interface as well as robotics and clinical care and treatment for illnesses like autism and sadness. Face expression analysis technology is examined in this dissertation, as well as the development of artificially intelligent systems for practical usage. In order to improve FER, a combination of geometric and textural domain feature extractors and several machine learning approaches is employed in the study's initial section. Once these two cutting-edge algorithms have been blended, a framework is built to test their individual and joint performance in the domain of facial recognition. Advanced facial alignment and localizing procedures are then employed to dissect the face into its constituent parts. Additionally, FER examines the use of Convolutional Neural Networks for deep learning (CNNs). Two proposed available facial expression databases are used to validate the proposed method. Using an ensemble neural network, the experimental findings indicate high accuracy for diverse datasets on the combined features.