Real-Time Heart Rate Monitoring Using Facial Recognition and Image Processing

Over the last few years, the union of computer vision with healthcare has paved the way for new possibilities of non-invasive monitoring systems as a more convenient and comfortable alternative to conventional systems. This paper suggests an original, real-time system for monitoring heart rate through facial recognition technologies. Through the use of image processing and machine learning, the system interprets the changes in facial features, such as skin color changes related to blood circulation, captured by a camera. A Haar Cascade Algorithm is employed for accurate facial detection, and areas related to heart rate variability are detected. The accuracy of the system was tested against gold-standard ECG signals with a mean absolute error (MAE) of 1.45 BPM in normal lighting and 2.85 BPM in movement conditions, with a root mean squared error (RMSE) of 3.05 BPM across conditions. The outcomes confirm the system's reliability for real-time use. By avoiding physical contact, the suggested solution provides greater user comfort, rendering it suitable for remote patient monitoring and telemedicine uses. This contribution addresses the expanding scope of smart healthcare by introducing a new, non-invasive approach to continuous vital sign monitoring.