Abstract: Non-invasive heart rate prediction has gained significant interest in various domains. Traditional methods utilizing external gadgets with the intention of of estimating heart rate through electrocardiogram (ECG)sensors often require direct skin contact, limiting comfort and utility. Recent developments in computer vision techniques have shown their capability to extract physiological data from facial videos through the detection of the forehead region The proposed real-time system employs OpenCV for facial recognition and tracking, ensuring with good lighting and minimal motion artifacts. Additionally, the study explores recognizing various physiological waveforms from raw data streams to enhance health monitoring capabilities. Shifting focus to image reports of ECGs, the research employs machine learning to digitize and analyses ECG paper records automatically. The transformation of ECG data into 1-D signals facilitates the extraction of P, QRS, and T waves, aiding in measuring heart electrical activity using diverse techniques. We employ dimension reduction techniques for feature extraction, and multiple classifiers such as ensemble, logistic regression, support vector machine (SVM), and k-nearest neighbors (KNN) are used for diagnosis. The resulting model demonstrates diagnostic potential, accurately identifying ECG records to interpret various cardiac conditions, such as myocardial infarction, arrhythmias, and normal heart function

| DOI: 10.17148/IARJSET.2023.10849