Real- Time Access Control System Based on Temperature Detection and the Correct Use of Face Masks

ABSTRACT

Nowadays, the COVID-19 pandemic has changed our lives. Some biosecurity measures have been implemented, including the use of face masks and the detection of body temperature;however, there are a lot of outbreaks, and the world cannot overcome this illness. In multiple cases, it is very difficult to measure the temperature and verify the correct use of face masks in everyone. Therefore, this paper proposes a real-time access control system based on body temperature detection and the correct use of face masks. This system uses a Raspberry Pi 4, which integrates temperature measurement using a thermal imager, the detection of the correct use of face masks using Convolutional Neural Networks (CNN), with a model built based on TensorFlow and MobileNetV2 that works on the video obtained from a thermographic camera using OpenCV and the Real Real-Time Streaming Protocol (RSTP). The system includes four modules body temperature detection, processes, access, and visual interface. As a result, the access control system establishes six classification cases high temperature and low temperature in faces without a mask, with an incorrectly placed mask, and with a correctly placed mask. The results show a system performance greater than 95% in all cases with a neural network model trained with a learning rate of 10E-4 and 15 epochs. © 2022 IEEE.