Abstract
Wearing safety helmets is a crucial preventive measure to minimize head injuries in industrial and construction environments preventing unauthorized entry into restricted zones and implementing regulations in helmet wearing are still remain significant challenges in workplace safety management. This work proposes a real-time helmet detection and alert system for the restricted zone based on advanced YOLOv8 developed to improve the accuracy detection. In this model, a Sparse Spatial Attention Mechanism (SSAM) is integrated into YOLOv8 to improve the network focus on required spatial regions by reducing the irrelevant additional features and utilize the modified slim-neck structure. It will be integrated with V0V-GSCSP and GSConv enables efficient multi-scale feature fusion with reduced complexity without the lack of accuracy. This advanced architectural model allows increased precision detection and make it suitable for real-time industrial applications. Additionally, this model improves the real-time alert system developed using python, FastAPI at the backend and Angular at frontend. The system provides the facility to connect an IP camera or web camera as an input source, allows the user to mark restricted zones visually and monitor detections in real time. When any person without a helmet enters the restricted area, an automatic alert message is generated using the Twilio API which instantly send alert message to registered phone number. The enhanced YOLOv8 model achieved better performance with precision of 93.49%, recall of 91.37% and F1-score of 92.40% for the helmet-on class and precision of 92.68%, recall of 91.89% and F1-score of 92.35% for helmet-no class. The overall performances were precision of 93.09%, recall of 91.63% and F1-score of 92.35%. The solution for automated industrial and construction site safety monitoring integrated with slim-neck, SSAM and optimization technique to increase the accuracy detection and reliability.
References
- Lin, Bingyan. "Safety helmet detection based on improved YOLOv8." IEEE Access 12 (2024): 28260-28272.
- Jin, Peijian, Hang Li, Weilong Yan, and Jinrong Xu. "Yolo-esca: a high-performance safety helmet standard wearing behavior detection model based on improved yolov5." IEEE Access 12 (2024): 23854-23868.
- Liu, Yiping, Benchi Jiang, Huan He, Zhijun Chen, and Zhenfa Xu. "Helmet wearing detection algorithm based on improved YOLOv5." Scientific reports 14, no. 1 (2024): 8768.
- Zhao, Yu, Quan Chen, Wengang Cao, Jie Yang, Jian Xiong, and Guan Gui. "Deep learning for risk detection and trajectory tracking at construction sites." IEEE Access 7 (2019): 30905-30912.
- Han, Kun, and Xiangdong Zeng. "Deep learning-based workers safety helmet wearing detection on construction sites using multi-scale features." Ieee Access 10 (2021): 718-729.
- Tran, Van Than, Thanh Sang To, Tan-No Nguyen, and Thanh Danh Tran. "Safety helmet detection at construction sites using YOLOv5 and YOLOR." In International Conference on Intelligence of Things, pp. 339-347. Cham: Springer International Publishing, 2022.
- Zhao, Baining, Haijuan Lan, Zhewen Niu, Huiling Zhu, Tong Qian, and Wenhu Tang. "Detection and location of safety protective wear in power substation operation using wear-enhanced YOLOv3 algorithm." IEEE Access 9 (2021): 125540-125549.
- Liu, Ruihao, Zhongxi Shao, Zhenzhong Yu, and Rui Li. "Research on real-time helmet detection and deployment based on an improved YOLOv7 network with channel pruning." Signal, Image and Video Processing 19, no. 2 (2025): 118.
- Wang, Lijun, Yunyu Cao, Song Wang, Xiaona Song, Shenfeng Zhang, Jianyong Zhang, and Jinxing Niu. "Investigation into recognition algorithm of helmet violation based on YOLOv5-CBAM-DCN." Ieee Access 10 (2022): 60622-60632.
- Li, Longlong, Zhifeng Wang, and Tingting Zhang. "Gbh-yolov5: Ghost convolution with bottleneckcsp and tiny target prediction head incorporating yolov5 for pv panel defect detection." Electronics 12, no. 3 (2023): 561.
- Park, Man-Woo, Nehad Elsafty, and Zhenhua Zhu. "Hardhat-wearing detection for enhancing on-site safety of construction workers." Journal of Construction Engineering and Management 141, no. 9 (2015): 04015024.
- Mneymneh, Bahaa Eddine, Mohamad Abbas, and Hiam Khoury. "Automated hardhat detection for construction safety applications." Procedia engineering 196 (2017): 895-902.
- Girshick, Ross, Jeff Donahue, Trevor Darrell, and Jitendra Malik. "Rich feature hierarchies for accurate object detection and semantic segmentation." In Proceedings of the IEEE conference on computer vision and pattern recognition, pp. 580-587. 2014..
- Girshick, Ross. "Fast r-cnn." In Proceedings of the IEEE international conference on computer vision, pp. 1440-1448. 2015.
- Ren, Shaoqing, Kaiming He, Ross Girshick, and Jian Sun. "Faster r-cnn: Towards real-time object detection with region proposal networks." Advances in neural information processing systems 28 (2015).
- Liu, Hai, Xiang Wang, Wei Zhang, Zhaoli Zhang, and You-Fu Li. "Infrared head pose estimation with multi-scales feature fusion on the IRHP database for human attention recognition." Neurocomputing 411 (2020): 510-520.
- Wang, Lili, Xinjie Zhang, and Hailu Yang. "Safety helmet wearing detection model based on improved YOLO-M." IEEE Access 11 (2023): 26247-26257.
- Li, Zhishan, Wenqing Xie, Lingzhi Zhang, Shan Lu, Lei Xie, Hongye Su, Weidong Du, and Weifeng Hou. "Toward efficient safety helmet detection based on YoloV5 with hierarchical positive sample selection and box density filtering." IEEE transactions on instrumentation and measurement 71 (2022): 1-14.
- Chen, Junhua, Sihao Deng, Ping Wang, Xueda Huang, and Yanfei Liu. "Lightweight helmet detection algorithm using an improved YOLOv4." Sensors 23, no. 3 (2023): 1256.
- Cheng, Rao, Xiaowei He, Zhonglong Zheng, and Zhentao Wang. "Multi-scale safety helmet detection based on SAS-YOLOv3-tiny." Applied Sciences 11, no. 8 (2021): 3652.