<![CDATA[Indonesia is one of the countries experiencing COVID-19 impacts. Various measures have been conducted to prevent the spread of this virus. One of the efficient measures to prevent this impact is by implementing a strict health protocol and proper mask-wearing. Mask-wearing monitoring continues to be carried out in office buildings, supermarkets, and other public spaces. The supervisor’s role is indispensable in supervising proper mask-wearing. However, a supervisor has limitations in conducting supervision, creating a gap for people not to comply with mask-wearing rules properly. Therefore, it is necessary to have a system that works automatically to assist supervisors in monitoring proper mask-wearing. This paper aims to design a computer vision capable of detecting whether or not a person wears a mask using the TensorFlow deep learning framework. TensorFlow is used for its efficiency in processing digital image data. The classification of digital image data in TensorFlow uses a Keras deep learning structure. As a result, it is lightweight and can be used on embedded devices such as Jetson Nano to detect mask-wearing in real time. The stages of a mask detection system consisted of image dataset collection, feature extraction, data separation, modeling, model training, and model implementation. TensorFlow deep learning framework processed image data directly through a webcam. When the camera captured the object of the person not wearing the mask properly, the monitor screen displayed a red box on the face. The sign can help the supervisor when conducting supervision. The test results show that the system successfully correctly detects unmasked people, with an accuracy of 99.48%. In addition, the system also managed to detect people wearing masks properly and got an average accuracy of 99.12%. The monitor displays a green box on the face when the detected person properly wears a mask.]]>
Copyrights © 2023
