Deep Learning-Based Automated Inspection of Generic Personal Protective Equipment

This study presents an automated system for monitoring Personal Protective Equipment (PPE) compliance using advanced computer vision techniques in industrial settings. Despite strict safety regulations, manual monitoring of PPE compliance remains inefficient and prone to human error, particularly in harsh environmental conditions like in Saudi Arabia’s Eastern Province. The proposed solution leverages the state-of-the-art YOLOv11 deep learning model to detect multiple safety equipment classes, including safety vests, hard hats, safety shoes, gloves, and their absence (no_hardhat, no_safety_vest, no_safety_shoes, no_gloves) along with person detection. The system is designed to perform real-time detection of safety gear while maintaining accuracy despite challenging conditions such as extreme heat, dust, and variable lighting. In this regard, a state-of-the-art augmented and rich dataset obtained from real-life CCTV, warehouse, and smartphone footage has been investigated using YOLOv11, the latest in its family. Preliminary testing indicates the highest detection rate of 98.6% across various environmental conditions, significantly improving workplace safety compliance and reducing the resources required for manual checks. Additionally, a user-friendly administrative interface provides immediate notification upon detection of breaches so that corrective action can be taken immediately. This initiative contributes to Industry 4.0 practice development and reinforces Saudi Vision 2030’s emphasis on workplace safety and technology.