AI-Based Intelligent Medical PPE and Equipment Detection System for Real-Time Healthcare Safety Monitoring

Abstract

Healthcare facilities must strictly enforce safety compliance, as these areas are extremely sensitive, particularly the ICU, OT and emergency wards. The conventional manual monitoring methodologies are ineffective, expensive, resource-hungry, and prone to human errors and associated issues. Thus, there are limitations in their use in large-scale, continuous monitoring. The intelligent monitoring system is AI-based monitoring application which helps in the detection of personal protective equipment (PPE), identifying health care personnel through facial biometrics, etc. The YOLOv8-based fast object detection, InsightFace-based face recognition and SORT-based multi-object tracking system tracks objects from frame to frame. This system uses React for the dashboard and FastAPI with MySQL at the backend as a full-stack app. The experimental results prove greater than 97.3%, accuracy, 96.8% precision, 95.1% recall, and an F1 score of 95.9% with an inference latency of 28 milliseconds per frame making it a suitable production for healthcare compliance applications.

Introduction

This paper presents the development of an AI-based automated PPE (Personal Protective Equipment) monitoring system designed to improve infection control and occupational safety in healthcare environments.

The study is motivated by the limitations of manual PPE compliance monitoring, which depends on human supervision and is often unreliable due to workload, fatigue, and staffing shortages. These issues became especially critical during COVID-19, where non-compliance significantly increased infection risks among healthcare workers. As a result, there is a strong need for automated, continuous, and scalable monitoring systems.

To address this, the authors propose a fully integrated computer vision system that combines:

• YOLOv8 for real-time PPE detection,
• InsightFace for facial recognition, and
• SORT for multi-object tracking.

Together, these components enable continuous identification and tracking of healthcare personnel while checking PPE compliance in real time. The system is designed as a production-ready full-stack web application capable of handling multiple camera feeds simultaneously.

The study highlights advancements in related technologies: YOLOv8 provides fast and accurate object detection, InsightFace delivers high-precision face recognition using deep metric learning, and SORT enables efficient real-time tracking of individuals across video frames. However, existing systems often treat detection, recognition, and tracking separately, creating a gap in integrated solutions.

To overcome this, the proposed system introduces a four-tier architecture:

1. Client layer (React-based dashboard for live monitoring and alerts),
2. Application layer (FastAPI backend handling authentication and API requests),
3. ML processing layer (AI models for detection, recognition, and tracking),
4. Data/streaming layer (managing video input and outputs).

The system also includes a custom healthcare dataset and optimized training methods to improve performance in real-world clinical environments.

Conclusion

Artificial intelligence-based intelligent monitoring for PPE compliance detection was presented in the paper. The system achieves 97.3% detection accuracy, 96.8% precision, 95.9% F1score, and real-time inference at 35.7 FPS on GPU hardware via a full-stack web application combining YOLOv8 for object detection, InsightFace for biometric facial recognition, and SORT for multi-object tracking. The compliance dashboard with an interactive module and automated creation of alerts along with a modular four tiers architecture make this system a handy and scalable option for the deployment hospitals, clinics, and other healthcare facilities with high risks. Future works will include robustness under difficult lighting conditions, application of appearance-aware tracking algorithms for dense crowded environments, increase the dataset diversity across different clinical sites and edge-computing deployment option to cut down the infrastructure cost. The proposed system significantly and practically advances the monitoring of healthcare safety compliance to fully automated AI.

References

[1] J. Redmon, S. Divvala, R. Girshick, and A. Farhadi, “You Only Look Once: Unified, Real-Time Object Detection,” Proc. IEEE CVPR, pp. 779–788, 2016. https://doi.org/10.1109/CVPR.2016.91 [2] A. Bochkovskiy, C.-Y. Wang, and H.-Y. M. Liao, “YOLOv4: Optimal Speed and Accuracy of Object Detection,” arXiv:2004.10934, 2020. https://doi.org/10.48550/arXiv.2004.10934 [3] Ultralytics, “YOLOv8: A State-of-the-Art Object Detection Model,” GitHub Repository, 2023. https://github.com/ultralytics/ ultralytics [4] J. Deng, J. Guo, N. Xue, and S. Zafeiriou, “ArcFace: Additive Angular Margin Loss for Deep Face Recognition,” Proc. IEEE/CVF CVPR, pp. 4690–4699, [5] 2019. https://doi.org/10.1109/ CVPR.2019.00482 [6] A. Bewley, Z. Ge, L. Ott, F. Ramos, and B. Upcroft, “Simple Online and Realtime Tracking,” Proc. IEEE ICIP, pp. 3464–3468, 2016. https://doi.org/10.1109/ICIP.2016.7533003 [7] World Health Organization, “Infection Prevention and Control: Healthcare Worker Safety and PPE Use,” WHO Technical Report, Geneva, 2023. https://www.who.int/publications/i/item/9789240078826 [8] R. Miikkulainen et al., “Deep Learning Applications in Healthcare Monitoring,” J. Healthcare Engineering, vol. 2024, 2024. https://doi.org/10.1155/2024/6645125 [9] M. Jocher et al., “Ultralytics YOLO: Real-Time Object Detection,”IEEE Trans. Medical Imaging, vol. 42, no. 6, pp. 1812–1824, 2023. https://doi.org/10.1109/TMI.2023.3245124 [10] [9] L. Zhang, R. Yang, and M. Shafiq, “PPE Detection in Healthcare: A Comprehensive Review,” Sensors, vol. 22, no. 4, p. 1346, 2022. https://doi.org/10.3390/s22041346 [10] M. Wang and W. Deng, “Deep Face Recognition: A Survey,” Neurocomputing, vol. 429, pp. 215–244, 2021. https://doi.org/10. 1016/j.neucom.2020.10.081 [11] I. Goodfellow, Y. Bengio, and A. Courville, Deep Learning. Cambridge, MA: MIT Press, 2016. ISBN: 978-0-262-03561-3. [12] O. Ronneberger, P. Fischer, and T. Brox, “U-Net: Convolutional Networks for Biomedical Image Segmentation,” Proc. MICCAI, pp. 234–241, 2015. https://doi.org/10.1007/978-3-319-24574-428 [13] N. Hasan, Y. Bao, A. Shawon, and Y. Huang, “AI Safety Monitoring in Clinical Environments,” IEEE Access, vol. 11, pp. 45623–45638, 2024. https://doi.org/10.1109/ACCESS.2024.3367890 [14] A. Kumar and S. Sharma, “Vision Systems Review for Medical Applications,” Expert Systems with Applications, vol. 237, p. 121562, 2025. https://doi.org/10.1016/j.eswa.2024.121562 [15] S. Ren, K. He, R. Girshick, and J. Sun, “Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks,” IEEE TPAMI, vol. 39, no. 6, pp. 1137–1149, 2017. https://doi. org/10.1109/TPAMI.2016.2577031 [16] H. Chen, F. Mao, and Z. Liu, “Healthcare AI Systems: Challenges and Opportunities,” npj Digital Medicine, vol. 7, no. 1, 2024. https://doi.org/10.1038/s41746-024-01063-1 [17] P. Cunningham, M. Cord, and S. Delany, “Supervised Learning,” in Machine Learning Techniques for Multimedia. Berlin: Springer, 2008. https://doi.org/10.1007/978-3-540-75171-7 2 [18] Y. LeCun, Y. Bengio, and G. Hinton, “Deep Learning,” Nature, vol. 521, no. 7553, pp. 436–444, 2015. https://doi.org/10.1038/ nature14539 [19] G. Litjens et al., “A Survey on Deep Learning in Medical Image Analysis,” Medical Image Analysis, vol. 42, pp. 60–88, 2017. https://doi.org/10.1016/j.media.2017.07.005 [20] A. Panesar, Machine Learning and AI for Healthcare, 2nd ed. Cham: Springer, 2021. https://doi.org/10.1007/978-1-4842-6537-6 [21] Y. Zhang et al., “ByteTrack: Multi-Object Tracking by Associating Every Detection Box,” Proc. ECCV, pp. 1–21, 2022. https://doi.org/10.1007/978-3-031-20047-2 1 [22] D. Shen, G. Wu, and H.-I. Suk, “Deep Learning in Medical Image Analysis,” Annual Review of Biomedical Engineering, vol. 19, pp. 221–248, 2017. https://doi.org/10.1146/ annurev-bioeng-071516-044442 [23] W. Liu, D. Lian, and Y. Zhou, “Automated PPE Detection Using Deep Learning,” Computers in Biology and Medicine, vol. 164, p. 107254, 2024. https://doi.org/10.1016/j.compbiomed.2023.107254 [24] F. Jiang and Y. Jiang, “Intelligent Healthcare AI: Recent Advances,” J. Medical Systems, vol. 49, no. 1, 2025. https://doi.org/10.1007/s10916-024-02128-z [25] S. Rajaraman, S. K. Antani, and G. R. Thoma, “AI-Based Monitoring in Critical Care,” BMC Medical Informatics and Decision Making, vol. 23, no. 1, p. 118, 2023. https://doi.org/10. 1186/s12911-023-02215-8

Copyright

Copyright © 2026 Mr. B. Murali Krishna, Sk. Saddam, U. Kusuma Sai Lakshmi, Ragolu Lakshmi Rupavathi, Vodugu Saranya, Abhiram Katari. 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.