Modeling and implementation of an automatic access control system for secure premises using facial recognition / Bopatriciat Boluma Mangata ... [et al.]

• Main Authors: Mangata, Bopatriciat Boluma, Mbambi, Kisiaka, Muamba, Kadima, Khalaba, Fundji
• Format: Article
• Language: English
• Published: Universiti Teknologi MARA, Perlis 2022
• Subjects: Programming. Rule-based programming. Backtrack programming; Detectors. Sensors. Sensor networks
• Online Access: https://ir.uitm.edu.my/id/eprint/68844/1/68844.pdf; https://ir.uitm.edu.my/id/eprint/68844/; https://crinn.conferencehunter.com/index.php/jcrinn

Security is a major concern within companies to prevent access to information by unauthorized persons. In this work, we are interested in access control through facial recognition. To realize this access control system based on facial recognition, we used an embedded system under Arduino which gives us the possibility to assemble the performances of programming and electronics, more precisely, we programmed electronic systems for the automatic opening of doors without the action of a human being. From a sample of 100 individuals composed of 40 women and 60 men, 75 of whom were registered and 25 non-registered, our access control system obtained the results of 70 true positives, 5 false negatives, 8 false positives and 17 true negatives that constitute our confusion matrix. However, from the set of tests performed we can conclude that multi-modality fusion can be leveraged to increase the performance of the verification system as the verification performance of multimodal systems (feature fusion or score fusion) can be applied to give even better results.