AUTOMATED DETECTION SYSTEM OF PERSONAL PROTECTIVE EQUIPMENT USING COMPUTER VISION IN A DELIMITED ENVIRONMENT

Abstract

This work presents an automated system for detecting Personal Protective Equipment (hard hat, safety glasses and boots) in industrial environments using computer vision. The YOLOv8n model was trained for 200 epochs and implemented on a Raspberry Pi 4 with a Coral Edge TPU, integrated with Firebase and Telegram. The methodology included real images, data augmentation techniques, and evaluation using standard detection metrics. The results showed 94.7% precision, 87.3% recall, and 93.9% mAP@0.5, with real-time performance of 5 FPS. It is concluded that the system is technically feasible for monitoring PPE as planned and shows that it may eventually provide effective support for occupational safety enforcement