Deteksi Kelengkapan Alat Pelindung Diri untuk Pekerja Proyek Menggunakan Metode You Only Look Once Versi 5 (YOLOv5)

Abstract

Toolset used for self-protection from any danger or accident that might happened in the environment. PPE consists of Hard Hat Helmet, Mask, Safety Vest, Boots. PPE is hard to monitored by Occupational Health and Safety (OHS) in every sector of the project environment at the same time. Because of labor and limited time, the workers took off their PPE as they like in the environment. Whereas in this case, it’s to minimalized fatal injury of themselves. With this limited time and the amount of labor, the application is needed to detect PPE. You Only Look Once Version 5 (YOLOv5) method will be used in this application to cover every sector of the environment. In this study, there are 7 classes which are Gloves, Helmet, Non-helmet, Person, Shoes, Vest, Bare-arms. The dataset consists of 1526 data which will be augmented and become 3958 images. The Data will be divided to 3628 training data, 295 validation data, and 29 testing data. The writer got the accuracy score of 92,07% in PPE detection. The result shows that YOLOv5 method is good in detecting PPE in workers body.