Vision Based PPE Detection Using Deep Learning

The goal of the personal protective equipment (PPE) detection at construction sites in this study is to employ contemporary deep learning technology to increase safety and compliance. Workers in the construction sector are frequently identified to dangerous situations, thus wearing PPE appropriately is crucial to their safety. However, manually keeping track of PPE usage takes a lot of effort, particularly on busy and large construction sites.This work suggests a method that automatically determines whether employees are donning the necessary PPE, such as Hard-hat ,vest and head using deep learning. The technology recognizes people images obtained from the construction sites and looks for PPE. A total of 17,248 training images, 2,455 test images, and 2,438 validation images obtained from Hard- Hat- Vest workers dataset from Kaggle platform are used to develop and validate the deep learning model. In this study YOLOV8 object detection model is employed to detect the PPEs such asHard-hat, vest and head. This fosters a culture of safety and accountability and helps avoid mishaps. AI not only improves monitoring effectiveness but also guarantees greater adherence to safety standards. For construction organizations seeking to use technology to lower hazards and increase workplace safety, this approach may be a useful tool.