Design and Development of a Laboratory Security System Using IoT-Based Face Recognition

An Internet of Things (IoT)-based laboratory security system is designed to improve security and monitoring efficiency. This study aims to develop such a system using ESP-32 CAM (Embedded Serial Peripheral Interface 32-bit Camera), a camera module with an ESP32 microcontroller with Wi-Fi connectivity and image processing capabilities for real-time access control and room monitoring. Firebase is a cloud-based facial data storage platform, and the Telegram application is a medium that allows automatic notifications. Access control testing was conducted involving five user samples. The system recorded a facial recognition success rate of 90% (9 out of 10 tests were successful), which is considered "good" based on the accuracy threshold of ≥ 85% for security applications. However, accuracy decreases in nonideal conditions, such as when the user's distance from the camera exceeds 30 cm or the angle of view of the face is not aligned with the camera. Non-ideal refers to operational situations where optimal parameters, such as distance and facial orientation, are unsatisfied. The notification delivery delay to the Telegram application was tested using Wireshark, with an average time of 18,528 ms, indicating that the system has a fast response in real-time. In addition, the monitoring camera (ESP-32 CAM) successfully sent stable real-time video to the laboratory manager's web without significant interference. The test results show that this system meets the design objectives as an efficient laboratory security solution. Some improvements are needed to improve accuracy under non-ideal conditions and expand monitoring testing to various network scenarios.