Construction Safety Detection: Using Python And Machine Learning

Abstract

Construction Safety Detection - Mail Alert (YOLOv8) is a real-time web-based platform designed to enhance safety on construction sites by automating the detection of safety gear such as helmets, vests, and masks worn by workers, and identifying the presence of individuals. Built using Python (for the YOLOv8 model), OpenCV, and various libraries for email notifications, the platform integrates real- time object detection with automated alerts. Utilizing the YOLOv8 algorithm, the system accurately detects whether workers are wearing appropriate safety equipment and provides real-time counts of safety gear and people present on the site. A key feature includes email alerts that are triggered when a worker is detected without the necessary gear, with a captured image of the incident. A non-blocking email process ensures that the video feed remains uninterrupted while alerts are sent in the background. The platform aims to improve construction site safety by providing quick notifications of potential safety hazards and fostering a proactive safety culture. The system resolves challenges such as API rate limits and high-resolution video processing by leveraging caching and asynchronous processing. Future enhancements include expanding the platform's functionality to support multiple construction site locations and integrating advanced analytics for safety trend predictions. This project addresses critical issues in construction site safety, improving compliance and reducing accident rates through automated and real-time monitoring.