Aplikasi Framework YOLO V.3 untuk Deteksi Pelanggaran Penggunaan Masker

Abstract

This research aims to develop a software that is able to detect violations of mask use such as wearing a mask on the chin or not covering the mouth and nose incompletely using YoloV.3. The software development method is the incremental method. This research is divided into two incremental stages. Incremental one focuses on developing a model that will be evaluated using confusion matrix. The result of incremental one is a model with a mAP value of 77.92%, precision 90.00%, recall 74.00%, and f1-score 81.00%. Incremental two focuses on building a software display which is then evaluated using the black box method. The final result of this research is software that can detect violations of mask usage.