PPE AND INDUSTRIAL SAFETY VIOLATION DETECTION USING YOLO

Abstract

Industrial workplaces such as manufacturing plants, construction sites, warehouses, and factories contain various safety hazards that can cause serious injuries or accidents. Personal Protective Equipment (PPE), including helmets, safety vests, gloves, safety shoes, and protective masks, plays an important role in reducing workplace risks. However, manually monitoring PPE compliance across large industrial areas is difficult and time-consuming. The PPE and Industrial Safety Violation Detection Using YOLO system is proposed as an automated computervision solution for detecting workers and identifying PPE-related safety violations in real time. The proposed system uses the YOLO (You Only Look Once) object-detection algorithm to analyze images and live video streams captured from CCTV or surveillance cameras. YOLO processes frames and identifies multiple objects simultaneously by generating bounding boxes, class labels, and confidence scores. The model can be trained to detect workers, helmets, safety vests, gloves, safety shoes, masks, and other safety-related objects depending on the requirements of the industrial environment. The system compares detected workers with the required PPE conditions and identifies violations such as a worker without a helmet, missing safety vest, absence of gloves, or failure to wear other required protective equipment. Each detected violation can be assigned a confidence score and severity level according to predefined safety rules. When a serious violation is identified, the system can generate a real-time alert for authorized safety personnel along with the camera location, timestamp, and visual evidence. A centralized dashboard can display live camera feeds, detected workers, PPE compliance rates, violation counts, severity levels, and historical safety statistics. The system can maintain records of detected violations for further analysis and reporting. Safety managers can use these records to identify frequently occurring violations, monitor safety compliance, and determine areas where additional training or corrective action may be required. Overall, the proposed system provides an efficient and scalable method for AIassisted industrial safety monitoring. By combining YOLO-based object detection, PPE verification, rule-based violation analysis, real-time alerts, and dashboard visualization, the system can reduce dependence on continuous manual observation. It is designed to support safety teams rather than replace human judgment, and future versions can incorporate additional capabilities such as restricted-area detection, fall detection, fire and smoke detection, worker-machine proximity monitoring, and advanced safety analytics.