Deep Learning-based Identification of Personal Protective Equipment in Construction Area
Work safety in the construction environment is highly dependent on the correct use of personal protective equipment (PPE). This study aims to develop an automatic PPE detection system using a YOLO-based deep learning model to improve supervision and compliance with PPE use in the field. Two variants...
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | Indonesian |
| Published: |
Islamic University of Indragiri
2025-07-01
|
| Series: | Sistemasi: Jurnal Sistem Informasi |
| Subjects: | |
| Online Access: | https://sistemasi.ftik.unisi.ac.id/index.php/stmsi/article/view/5271 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Work safety in the construction environment is highly dependent on the correct use of personal protective equipment (PPE). This study aims to develop an automatic PPE detection system using a YOLO-based deep learning model to improve supervision and compliance with PPE use in the field. Two variants of the YOLO model, namely YOLOv10 and YOLOv11, were tested and their performance was compared through a fine-tuning process using custom dataset consisting of 16,568 annotated images of construction workers wearing various types of PPE. The model was evaluated using precision, recall, and mAP50. The results showed that the YOLOv11s model performed the best with with an mAP50 of 0.718 and a precision score of 0.804, indicating good detection and classification ability. This model is able to detect various types of PPE effectively, so it can be used as a tool in real-time occupational safety monitoring. This study proves that the application of YOLO-based deep learning technology can be an effective solution to improve compliance with PPE use and reduce the risk of work accidents in the construction sector. The implications of this study open up opportunities for the development of more sophisticated and adaptive automatic monitoring systems in the future such as deploying the model on edge devices for real-time inference and expanding detection capabilities to include additional safety violations such as the absence of safety harnesses or proximity to hazardous zones. |
|---|---|
| ISSN: | 2302-8149 2540-9719 |