Design Integrated Staff Welcoming and Administration System Based on Facial Recognition for Smart University

In recent years, facial recognition technology is not only applied for security, healthcare, business but it is also exploited creatively in education and for smart university development. In this study, an Automated Employee Attendance Management System has been designed and implemented in a university's building for welcoming staff and supporting university administration. The system starts with a facial detection module which is based on the pre-trained MultiTask Cascaded Convolutional Network model. Then the feature vectors are created by using the ResNet34 network which results in the 128-dimension embedding vector. Face recognition is carried out by using various techniques such as K-Nearest Neighbors, Support Vector Machine. Besides the welcoming front-end module, a web-based application for querying information to manage people entering the building is also built as a back-end module. The proposed face recognition models have been trained and tested on a collected face database and a self-built face database of employees who work at the university building. The evaluation results show high recognition rates in terms of Precision, Recall, Accuracy, Fl-score and reasonable processing time. The proposed system has been piloted at the university for further development and research on face recognition technologies and smart building management.