PERSONAL PROTCTIVE EQUIPMENT DETECTION FOR OCCUPATIONAL SAFETY AND HEALTH USING YOLOV8 IN MANUFACTURING COMPANIES

According to data from BPJS Keltelnagakelrjaan, 265,333 cases of work accidents were recorded in 2022. The use of personal protective equipment (PPE) is very important in reducing and preventing work accidents in the company. Although PPE cannot eliminate all risks, it is possible to minimise the number of work accidents in manufacturing companies. The aim of this research is to automatically select Personal Protective Equipment (PPE) in the form of hard hats and vests and to improve the accuracy results using the YOLOv8 model. With a dataset of 500 helmet and velst images for deltelksi which will be categorised into 4 classes namely hellelm, velst, no-hellelm, no-velst. The dataset used is 500 data, which is then divided into three datasets, namely: training data as much as 70%, validation data as much as 20%, and telst data as much as 10%, from the dataselt telrselbut the best results of testing data values from 50 dataselt the accuracy results obtained are 0.98. It is hoped that with the use of Meltode and accuracy results using Yolo v8, it can be used in companies by detecting Personal Protective Equipment (PPE) with fast and accurate results, so that it can be applied in monitoring the use of PPE in manufacturing companies to reduce the risk of work accidents in manufacturing companies