Safe Sight: AI-Based Multi-Industry PPE Detection System Using YOLOv8

Abstract

Workplace safety is a major concern across industries such as construction, mining, pharmaceuticals, food packaging, and healthcare, where compliance with Personal Protective Equipment (PPE) standards is essential to reducing occupational hazards. Traditional monitoring methods rely heavily on manual supervision, which is often error-prone, inefficient, and difficult to scale in dynamic environments. This paper presents Safe Sight, an AI-based multi-industry PPE detection system built on the YOLOv8 object detection framework. The model was trained on a diverse dataset covering six critical PPE classes: helmet, face mask, safety vest, gloves, goggles, and surgical gown. Experimental evaluation demonstrated strong performance in terms of mean Average Precision (mAP), precision, recall, and F1-score, validating the model’s effectiveness for real-time applications. The system was deployed in a Python-based PyCharm application, enabling video and webcam-based detection with clear compliance indicators—green for PPE present and red for missing PPE. While the current prototype focuses on detection, the framework is scalable and can be extended with industry-specific datasets and IoT-based alerting and reporting systems to further strengthen workplace safety management.