Development of an Improved Convolutional Neural Network for an Automated Face-Based University Attendance System

Abstract

Because of the flaws of the present university attendance system, which has always been time intensive, not accurate, and a hard process to follow. It, therefore, becomes imperative to eradicate or minimize the deficiencies identified in the archaic method. The identification of human face systems has evolved into a significant element in autonomous attendance-taking systems due to their ease of adoption and dependable and polite engagement. Face recognition technology has drastically altered the field of Convolution Neural Networks (CNN) however it has challenges of high computing costs for analyzing information and determining the best specifications (design) for each problem. Thus, this study aims to enhance CNN’s performance using Genetic Algorithm (GA) for an automated face-based University attendance system. The improved face recognition accuracy with CNN-GA got 96.49% while the face recognition accuracy with CNN got 92.54%.