The most significant indicator of cardiovascular disease is a rapid heart rate. Traditional methods for detecting heart rate include the use of Electrocardiogram (ECG) and oximeter sensors. However, ECG requires skin contact with gel patches, which can be irritating for patients. Conversely, when pulse oximeter sensors are applied to the fingertip or earlobes for extended periods, they may lead to discomfort. Heart rate serves as a critical physiological metric for assessing heart function and overall physical health. It also plays a vital role as a medical indicator for diagnosing cardiovascular disorders and certain chronic conditions. The number of heart beats per minute, commonly known as heart rate, is an important physiological indicator that can reveal psychological and physical conditions. Remote vital sign sensing employs cam- eras instead of skin-contact sensors to provide improved access to measuring conditions. This approach relies on the principles of color and motion. Remote photoplethysmography (rPPG) examines changes in the tone of the skin layer of the face due to blood circulation, while remote ballistocardiography (rBCG) analyzes the minute head movements induced by the heartbeat. In this study, we present a machine learning strategy aimed at enhancing the accuracy of heart rate detection in naturalistic measurements.