A Contactless Visitor Access Monitoring System

Citation

Bong, Yi-Sin and Lee, Gin Chong (2021) A Contactless Visitor Access Monitoring System. International Journal on Robotics, Automation and Sciences, 3. pp. 33-41. ISSN 2682-860X

[img] Text
A Contactless Visitor Access Monitoring System.pdf
Restricted to Repository staff only

Download (597kB)

Abstract

This project presents a contactless visitor access monitoring in small premises which implemented deep learning model in face recognition, develop the graphical user interface (GUI) for new visitor registration and visitor identification. Five stages of monitoring process are designed in the contactless visitor access monitoring (CVAM) GUI, the first step is to give instructions to the admin user regarding the monitoring process, the second step is to perform face recognition, the third step is to scan the body temperature, the fourth step is to perform mask detection on the visitor, and the final stage is to record visitor access time. Another visitor registration (VisReg) GUI is designed to register new visitors into the system. In VisReg, admin user is required to pre-process face images with MTCNN technique and generate new classifier with a ResNet pre-trained model. The contactless visitor access monitoring process is demonstrated. The face recognition gives an accuracy of 82%, while the mask detection gives an accuracy of 95% when tested with the validation dataset. It can be concluded that the visitor monitoring process can be carried out in a contactless way to eliminate the close contact between the security officers, receptionist, and visitors.

Item Type: Article
Uncontrolled Keywords: Deep Learning, Face Recognition, Mask Detection, Contactless, Visitor Monitoring
Subjects: Q Science > Q Science (General) > Q300-390 Cybernetics
Divisions: Faculty of Engineering and Technology (FET)
Depositing User: Ms Nurul Iqtiani Ahmad
Date Deposited: 06 Dec 2021 04:59
Last Modified: 06 Dec 2021 04:59
URII: http://shdl.mmu.edu.my/id/eprint/9797

Downloads

Downloads per month over past year

View ItemEdit (login required)