This research presents the design and evaluation of an automatic locker system employing Local Binary Pattern Histogram (LBPH)-based facial recognition for secure, keyless access. The system integrates a webcam, Arduino Uno, relay module, and solenoid door lock, managed via a laptop interface. Performance was assessed through single-user variability testing under different facial accessories (66.7% recognition rate) and multi-user discrimination testing (50% accuracy across diverse profiles). Results indicate LBPH’s robustness against moderate occlusions but a notable decline with heavy facial coverage like helmets and masks. These findings emphasize the critical role of facial visibility in biometric reliability, offering practical insights for real-world security applications.
Copyrights © 2026