Data-Driven Contactless Health Monitoring System
Our focus on contactless health monitoring underscores its significant benefits, offering a safer and more accessible approach to healthcare. In our research, we're engaging in two distinct projects aimed at harnessing the potential of off-the-shelf cameras for regular health monitoring purposes. Firstly, our efforts revolve around remote photoplethysmography (rPPG), a technique through which we endeavor to extract heart rate data from skin videos captured by these cameras. By leveraging advanced algorithms, our goal is to accurately discern heart rate information, enabling non-invasive monitoring that can be seamlessly integrated into everyday life. Secondly, our research extends to breathing rate (BR) estimation, where we aim to analyze video data to infer breathing patterns. This entails detecting subtle movements indicative of breathing activity, thereby enabling the estimation of breathing rates without the need for physical sensors. Through this approach, we aim to develop a comprehensive monitoring system that provides valuable insights into two of the human vitals (heart and breathing rates), using readily available camera technology.
Figure (a) Developed RhythmEdge solution to estimate HR from video data using resource constraints devices.
Figure (b) Self-supervised approaches to learn robust rPPG function from noisy rPPG-data labels.
Figure (c): Breathing rate from video data.
Team
Principle Investigator
Nirmalya Roy
Research Assistant Professor
Anuradha Ravi
Ph.D Students
Zahid Hasan
Publications
Hasan, Zahid, Abu Zaher Md Faridee, Masud Ahmed, Shibi Ayyanar, and Nirmalya Roy. "SrPPG: Semi-Supervised Adversarial Learning for Remote Photoplethysmography with Noisy Data." In 2023 IEEE International Conference on Smart Computing (SMARTCOMP), pp. 25-32. IEEE, 2023. (best paper award).
Acknowledgements
This material is based upon work supported by the National Science Foundation under Grant Numbers: 1750936 (University of Maryland Baltimore County)