Face Recognition Attendance System

Abstract

Automated, reliable and contactless attendance systems have been evolving in the educational institutions and organizations. The attendance management based on face recognition has gained the importance. Attendance systems that are traditional may require manual attendance taking, are tedious and prone to error, attendance by proxy, and maintenance of incorrect attendance. To overcome these restrictions an intelligent attendance tracking system based on facial recognition techniques has been created. The method is based on a labelled database of facial images captured with multiple subjects in different environments and facial expressions. The data set was analysed to understand the distribution and quality of images, followed by a process of preprocessing the images such as normalisation, scaling, feature extraction, and organising. Processed data were split into training and testing data for model validation purposes. Three classification methods are used for training and application of the face identification, they are KNN, SVM, CNN. Three metrics were used to evaluate the performance: accuracy, precision, recall and F1-score. Results of the experiment show that the SVM model gives the best result with an accuracy of 98.0%, precision of 98.33%, recall of 98.0% and F1-score of 97.98%, outperforming other models. The framework was linked with a Flask-based interface to provide user registration, authentication, attendance marking and result visualisation which leads to improved efficiency and reliability in attendance recording.