Robust Facial Recognition System

Abstract

Access control and security are two areas where facial recognition technology is being used more and more. A robust facial recognition system that is able to recognize faces with a mask is a needed upgrade from its predecessor which can only scan faces without masks. This is due to the current pandemic where most places require their employees to wear masks when attending to work. This especially complicates the attendance clock process when the employees would have to remove their masks to have their faces scanned. This can ultimately increase the transmission risks which can endanger the entire workplace. With this system installed on the premises, employees can easily scan their faces to clock in their attendance without having to remove their masks which can also reduce any crowding. In this project, the ESP32 camera module, Arduino IDE, and Python were used to create the facial recognition safety system with a mask. Any visitors or staff are photographed and recorded on video by the system, which then uses machine learning algorithms to identify them using facial recognition and logs their information in an Excel spreadsheet. In a fully functioning system, the system will be able to scan and recognize an individual’s face when they are standing in front of the camera with average accuracy, precision, sensitivity, and specificity scores are given 76.92, 85.26, 83.51, and 57.57 respectively. If the individual has already been registered in the system, then the system will be able to detect the individual and their log-in or clock-in details will be recorded in the attendance sheet. This makes the system can be implemented in any facility since it is simple to operate, affordable, and has a compact footprint.