Smart Access Control System Based On Uncontrolled Environment Human Face Recognition Using Convolutional Neural Network

Abstract

Neural networks or other artificial intelligence methods have developed rapidly over the past four decades. Its use in various fields makes many people compete to develop it further. One of the applications of artificial intelligence is an automatic door opening and closing system. This development can provide many advantages for users, one of which is that there is no need to make direct contact with the door handle. Armed with a capable PC and an esp32 microcontroller, the system works by detecting images of user facial expressions approaching the object using a webcam. If the required input matches the system rules, the motor will move to open the door. By using convolutional neural network technique, the system can classify the image quickly. Several expressions such as angry, disgusted, scared, happy, sad, surprised, and normal can be the door-opening key of the system. The user can select one to use as the input key to drive the motor to open the door. The study outcomes for several predetermined facial expressions yielded an accuracy rate of 60% and a detection time of under 4 seconds. The detectable distance extends to ± 2 meters. Further study could enable the development of this autonomous door with an IoT-based system for enhanced efficiency. Hopefully, this research can influence the development of intelligent building systems and other fields of artificial intelligence technology.