<![CDATA[The increasing demand for secure and intelligent storage systems has encouraged the development of biometric-based authentication technologies. This study presents the design and implementation of a smart locker security system based on facial recognition using the YOLO (You Only Look Once) algorithm integrated with an ESP32-CAM module and IoT communication. The system was developed to provide automated user authentication and remote locker control through real-time face detection and recognition. Performance evaluation was conducted by analyzing face detection accuracy, authentication response time, and network communication quality under different distances and lighting conditions. Experimental results showed that the YOLO-based system achieved a a face detection success rate of up to 92% under both bright-light and low-light environments at distances of 30 cm and 100 cm. The detection error ranged from 2.60% to 4.30%, indicating stable and reliable detection performance. Authentication testing revealed that the system performed optimally at 30 cm, with response times ranging from 1.47 s to 2.56 s and error rates below 1.2%. At 100 cm, the response time increased to approximately 9–10 s, accompanied by higher error rates due to reduced facial feature visibility. In addition, network performance evaluation based on latency, packet loss, and throughput demonstrated reliable Wi-Fi communication between the ESP32-CAM and the server, ensuring smooth data transmission during authentication and locker control operations. The results confirm that the YOLO algorithm is effective for real-time facial recognition applications and can be successfully implemented in smart locker security systems. The proposed system provides accurate authentication, reliable communication, and enhanced security, making it suitable for practical access-control applications.]]>
Copyrights © 2026
