DETECTION OF PERSONAL PROTECTIVE EQUIPMENT IN EXTREME CONSTRUCTION CONDITIONS USING ANN

Abstract

The number of deaths in the construction industry is greater than in other industries through a number of countermeasures.  Although workers may intentionally or unintentionally neglect to wear such safety measures, Personal protective equipment (PPE) was continuously being developed to prevent this types of accidents. Performing a safety check manually might be difficult since there can be a lot of coworkers at a site. It is essential to identify worker noncompliance with PPE in an automated and real-time manner. Detection of Personal Protective Equipment in Extreme Construction Conditions Using ANN is the topic of this paper. The web-based collection of 2,509 images from video recordings of many construction sites are utilized as the model's training data set. This Artificial Neural Networks (ANN) model is utilised in the study, which makes use of transfer learning and a basic variation of the YOLOv5 deep learning network. A dataset called CHVG to identify the workers PPE. Described model achieves the parameters as Accuracy as 97%, Recall 97% and Precision 96%. Overall, the analysis shows that computer vision-based techniques for automating safety-related compliance processes on construction sites are both feasible and useful.