Development of YOLOv5-Based Real-Time Smart Monitoring System for Increasing Lab Safety Awareness in Educational...

Area of Science:

• Computer Vision
• Laboratory Safety
• Artificial Intelligence

Background:

• Laboratory safety is critical in educational settings, requiring students to use personal protective equipment (PPE).
• Current methods for monitoring PPE compliance are manual, time-consuming, and often ineffective.
• Unsupervised unsafe acts can lead to accidents in laboratories.

Purpose of the Study:

• To propose a real-time, smart vision-based system for monitoring student PPE compliance in educational labs.
• To automatically verify the use of masks, gloves, lab coats, and goggles using image/video analysis.
• To enhance laboratory safety and foster a health and safety culture among students.

Main Methods:

• Developed a novel dataset, Students Lab Safety (SLS), comprising 481 images of PPE (gloves, helmets, masks, goggles).
• Trained and compared YOLOv5 (YOLOv5l, YOLOv5m, YOLOv5n, YOLOv5s, YOLOv5x) and YOLOv7 models for PPE detection.
• Evaluated model performance using precision, F1 score, recall, and mean average precision (mAP).

Main Results:

• All trained models demonstrated promising performance in detecting PPE in educational laboratory settings.
• The YOLOv5n model achieved the highest mAP of 77.40% for both small and large instances.
• YOLOv5m followed with an mAP of 75.30%, indicating effective detection capabilities.

Conclusions:

• The proposed smart vision-based system effectively monitors student PPE compliance in real-time.
• This technology can automate safety checks, reduce manual monitoring efforts, and improve overall lab safety.
• The system aids in creating smarter laboratories and establishing a robust health and safety culture.