Development of Machine Learning Model for Detection of Ppe Kit

Abstract

Proper monitoring and detection of personal protection equipment is important for assuring societal health and safety, especially during emergencies concerning the health and wellbeing of the public. The aim of the research is to detect PPE using Detectron2 model in healthcare industry. This research concentrates on five classifications namely, goggles (safety glass), mask (face mask), face shield (helmet), coverall (vest/gown) and gloves. The research developed custom Detectron2 model and adopted the FAIR’s base Detectron2 model to identify the PPE suits. The proposed custom model includes additional layers to the ResNet (backbone) architecture to heighten the performance of the model. From the outcomes of the research, classification accuracy of the model is identified to that of 92%.