Facial Health Monitoring

Non-contact facial health monitoring has emerged as an effective alternative to conventional sensor-based systems for continuous and remote healthcare applications. This paper presents an AI-powered facial health monitoring system that estimates vital physiological parameters such as heart rate (HR), blood pressure (BP), oxygen saturation (SpO₂), emotion, and stress level from live facial video input. The proposed approach employs camera-based face detection, preprocessing, data cleaning, facial feature extraction, and optimized remote photoplethysmography (rPPG) signal analysis. To improve robustness against motion artifacts, illumination variations, and noise, a Kalman Filter is applied for real-time signal smoothing and stabilization. Experimental results obtained from multiple users under different lighting and posture conditions demonstrate reliable and consistent performance. The system successfully detected faces in real time and estimated HR values in the range of 56.4–65.4 BPM, SpO₂ consistently around 99.9%, and BP values ranging from 151/98 mmHg to 170/106 mmHg. Additionally, emotion was identified as Neutral (30.0%) with stress levels classified as Relaxed or Normal, indicating stable physiological and mental states. The smooth and continuous outputs after Kalman filtering confirm effective noise reduction and signal stability. The results validate that the proposed system enables accurate, real-time, and contactless health monitoring, making it suitable for smart healthcare and remote patient-monitoring environments.