Deep Learning Framework for Automated Worker Helmet Detection And Safety Compliance Monitoring Using The Yolo Object Detection Model

Abstract

Construction automation tools that implement computer vision technologies like “YOLO and Re-CNNs actively contribute to enhancing safety at construction sites with their automated monitoring and detection systems. The system presented in this paper harnesses the YOLO detection model, and utilizes the Flask web application to detect in real time if construction workers are wearing safety helmets, and to check if construction safety compliance is being upheld. This framework offers continuous automated monitoring of construction safety compliance in lieu of the traditional manual check systems, and therefore reduces both systematic and human errors. The system incorporates an automated notification system to alert construction safety compliance officers to the occurrence of a safety compliance violation in order to facilitate on the spot corrective actions. The Flask web application offers an easy-to-use Upload interface for video files and real time web cam control with no technical skills required of the user. The system is designed to record safety compliance violations for future reviews and to facilitate safety compliance violation reporting. The built-in YOLO detection model reportedly offers powerful detection capabilities in a variety of different environments, in addition to being able to handle changes in visibility and abrupt occlusions. The combination of deep learning, computer vision, and web technologies offers a highly effective system to ensure compliance with construction safety regulations.