Enhancing Construction Worker Safety through YOLOv8-Based PPE Detection and Sensor-Driven Smart Helmets

Abstract

In order to improve worker safety in construction settings, this study proposes an integrated safety system that combines real-time sensor monitoring with deep learning. To ensure adherence to PPE regulations, the first component uses the YOLOv8 algorithm for precise and effective helmet detection. In order to monitor motion and environmental threats and send real-time alerts, the second component includes a Smart Helmet that is outfitted with a MEMS sensor, gas sensor, DHT11, and GSM module. When combined, these modules provide enhanced on-site safety and a proactive approach to accident avoidance.

Introduction

1. Overview

Workers in hazardous environments (mining, construction, gas plants) face risks such as toxic exposure, extreme temperatures, and physical injuries. Ensuring the proper and consistent use of Personal Protective Equipment (PPE) is critical but challenging due to manual monitoring limitations.

2. Proposed Solution

The project introduces a dual-layer safety system combining:

• YOLOv8-based deep learning model for real-time visual detection of PPE (helmets, vests).

• A smart helmet embedded with environmental and motion sensors for hazard monitoring and alerting.

3. YOLOv8 PPE Detection System

• Model: YOLOv8 architecture trained using helmet/no-helmet image datasets.

• Enhancements: Transfer learning, data augmentation, coordinate attention, and lightweight modules (Ghostv2).

• Performance: High detection accuracy across varied conditions.

  • Helmet: 96.5% precision, 95.2% recall

  • Vest: 94.3% precision, 92.7% recall

  • Mask: 93.1% precision, 91.8% recall

  • Person: 97.8% precision, 96.9% recall

  • Machinery: 90.4% precision, 88.6% recall

4. Smart Helmet System

• Hardware: Arduino-based helmet with sensors

  • MQ137 Gas Sensor: Detects toxic gases

  • DHT11: Monitors temperature and humidity

  • MEMS Accelerometer: Detects falls and abnormal head motion

• Communication & Alerts:

  • GSM Module: Sends SMS alerts to safety personnel

  • Buzzer: Provides instant on-site warnings

• Functionality:

  • Alerts triggered for heat stress, toxic gas presence, or sudden impacts

  • Alerts are generated within 2–5 seconds

5. Literature Review Insights

• Previous studies used earlier YOLO versions (e.g., YOLOv3) and basic smart helmet prototypes.

• YOLOv8 offers improvements in detection under occlusion and dynamic conditions.

• Hybrid systems integrating vision and sensor data show promise but face challenges in scalability and real-time deployment.

6. Methodology

• Training and testing done using pre-collected datasets under simulated conditions.

• YOLOv8 refined for real-time PPE detection.

• Smart helmet sensors calibrated for threshold-based alerting in hazardous conditions.

7. Key Findings

• The integrated system offers fast, accurate, and automated PPE compliance monitoring.

• Sensor module effectively detects environmental and physiological hazards.

• Alerts (audible and SMS) are timely, improving emergency response capability.

• The solution reduces dependence on manual inspection and enhances safety in high-risk workplaces.

Conclusion

In conclusion, by automating helmet recognition, the Helmet recognition System built with YOLO V8 offers a strong, real-time way to improve safety on building sites. By combining computer vision and machine learning techniques, the system provides a dependable and effective means of ensuring that all employees are wearing helmets, so enhancing workplace safety. The system can quickly and precisely identify helmets in photos by utilizing YOLO V8\'s object detection capabilities. Users can easily input pictures and get real-time helmet compliance forecasts because to the combination of a solid Python backend and an easy-to-use, interactive Streamlit user interface. This automated detection system reduces human error and raises overall safety standards at construction sites by doing away with the need for manual safety checks and enabling safety managers to monitor compliance in a more scalable and economical way.To guarantee a seamless user experience, the system\'s backend manages crucial tasks like image preprocessing, model training, and prediction serving. After being trained on a collection of labelled photos, the YOLO V8 model produces predictions quickly and accurately, and the frontend gives users unambiguous visual feedback in the form of confidence scores and bounding boxes. Users can easily upload photographs, evaluate findings, and comprehend real-time predictions because to the system\'s dynamic and engaging interface, despite its simplicity. More sophisticated functions can be added to the system in the future, like extending its functionality to identify additional safety gear like gloves or vests and connecting it with security cameras for real-time helmet monitoring. The technology can play a vital role in safety enforcement and accident prevention across a variety of domains, with potential applications in industries other than construction.There are a number of ways to expand and improve the Helmet Detection System\'s capabilities as it develops. First, by integrating live video feeds, building sites might identify helmets in real time, instantly alerting safety inspectors when a worker is discovered not wearing one. To further guarantee worker safety, the system should be expanded to detect additional personal protective equipment (PPE), such as boots, gloves, and safety vests. A cloud-based storage function for pictures and forecasts might be added to the system, allowing safety managers to examine past data and monitor changes in helmet compliance over time. Furthermore, incorporating more sophisticated machine learning models and methods—like multi-task learning or transfer learning—could increase the speed and accuracy of detection.

References

[1] M. S. Arshad, M. A. Khan, and S. U. Jan, “Smart Helmet for Hazard Detection and Worker Safety Monitoring using IoT,” International Journal of Computer Applications, vol. 182, no. 17, pp. 25–29, 2018. [2] Sharma and R. K. Gupta, “Automatic Safety Helmet Wearing Detection,” International Journal of Computer Vision and Image Processing (IJCVIP), vol. 9, no. 1, pp. 60–76, 2019. [3] J. Patel, R. Patel, and H. Shah, “PPE Compliance Monitoring Using YOLO Object Detection Algorithm,” International Journal of Innovative Research in Science, Engineering and Technology (IJIRSET), vol. 9, no. 6, pp. 5568–5574, 2020. [4] N. Jain, M. K. Patel, and K. Shah, “Real-time Personal Protective Equipment (PPE) Detection using Deep Learning,” Proceedings of the International Conference on Intelligent Computing and Control Systems (ICICCS), 2021. [5] M. T. Hasan, S. R. U. Ahmed, and M. S. Hossain, “A Smart Safety Helmet Using IoT and Sensor Technology for Worker Protection in Hazardous Areas,” IEEE International Conference on Robotics, Automation, Artificial Intelligence and Internet-of-Things (RAAIIoT), 2022. [6] Ultralytics, “YOLOv8 - Real-Time Object Detection and Fine Tuning,” GitHub Repository, 2023. [Online]. Available: https://github.com/ultralytics/ultralytics [7] Roboflow, “Roboflow Dataset and Annotation Tool,” 2023. [Online]. Available: https://roboflow.com [8] H. S. Lee, Y. H. Park, and J. W. Jung, “Development of a Construction Worker Safety Monitoring System Using Wearable Smart Helmet and Sensors,” Automation in Construction, vol. 122, pp. 103–113, 2021. [9] \"Smart helmet with real-time accident detection and tracking,\" article by S. K. Sharma and R. Jindal, International Journal of Advanced Research in Computer Science and Software Engineering, vol. 5, no. 4, 2015, pp. 234–238. [10] G. Z. A. Rajkumar and R. P. S. Nair, \"Smart helmet system with fall detection and safety alerts using machine learning,\" Procedia Computer Science, vol. 132, pp. 243-248, 2018. [11] P. G. Avasarala, R. S. R. Choudary, and D. S. Rajesh, \"IoT-based smart helmet system for monitoring health and safety of workers,\" Procedia Computer Science, vol. 132, pp. 57-63, 2018. [12] H. Patel, A. A. Patel, and R. S. Bhalala, \"Design and implementation of smart helmet with health monitoring system,\" International Journal of Advanced Research in Electronics and Communication Engineering, vol. 5, no. 4, pp. 54-59, 2016. [13] Z. Ye, C. Feng, and S. Jiang, “CIB-SE-YOLOv8: An Optimized YOLOv8 Model for Real-Time Safety Equipment Detection on Construction Sites,” arXiv preprint arXiv:2410.20699, 2024. [Online]. Available: https://arxiv.org/abs/2410.20699 [14] T. Wang, X. Liu, and J. Zhang, “Detection Method for Safety Helmet Wearing on Construction Sites Based on UAV Images and YOLOv8,” Buildings, vol. 15, no. 3, p. 354, 2024. [Online]. Available: https://www.mdpi.com/2075-5309/15/3/354 [15] P. R. Gokhale, P. Khandagale, and M. B. Potdar, “Real-Time Personal Protective Equipment Non-Compliance Recognition on AI Edge Cameras,” Electronics, vol. 13, no. 15, p. 2990, 2024. [Online]. Available: https://www.mdpi.com/2079-9292/13/15/2990 [16] B. Patel and M. Joshi, “Personal Protective Equipment Detection: A Deep-Learning-Based Sustainable Approach,” Sustainability, vol. 15, no. 18, p. 13990, 2023. [Online]. Available: https://www.mdpi.com/2071-1050/15/18/13990 [17] X. Liu et al., “CIB-SE-YOLOv8: An Optimized YOLOv8 Model for Real-Time Safety Equipment Detection on Construction Sites,” arXiv preprint arXiv:2410.20699, Oct. 2024. [Online]. Available: https://arxiv.org/abs/2410.20699 [18] T. Wang et al., “Detection Method for Safety Helmet Wearing on Construction Sites Based on UAV Images and YOLOv8,” Buildings, vol. 15, no. 3, p. 354, 2024. [Online]. Available: https://www.mdpi.com/2075-5309/15/3/354 [19] P. R. Gokhale et al., “Real-Time Personal Protective Equipment Non-Compliance Recognition on AI Edge Cameras,” Electronics, vol. 13, no. 15, p. 2990, 2024. [Online]. Available: https://www.mdpi.com/2079-9292/13/15/2990 [20] B. Patel and M. Joshi, “Personal Protective Equipment Detection: A Deep-Learning-Based Sustainable Approach,” Sustainability, vol. 15, no. 18, p. 13990, 2023. [Online]. Available: https://www.mdpi.com/2071-1050/15/18/13990

Copyright

Copyright © 2025 Mr. B. Dillibabu, Mr. A. Jyothinadh, Mr. D. Bavajan, Mr. D. Puneeth Kumar, Dr. R. Karunia Krishnapriya, Mr. Pandreti Praveen, Mr. V Shaik Mohammad Shahil, Mr. N. Vijaya Kumar. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.