Real-Time Detection of Human Heart Rate and Blood Pressure During Exercise

Regular moderate exercise has great benefits for both our physical and mental well-being. Nowadays, more and more people attach great attention to exercise, and among the population that needs to engage in moderate exercise, individuals with hypertension have a high proportion. Although there has been great progress in non-contact detection of heart rate and blood pressure, little research has been done to achieve detection during exercise. Different from the traditional contact measurement and IPPG (Imaging Photo Plethysmography)-based static heart rate and blood pressure detection, this paper designs a non-contact system that can monitor heart rate and blood pressure in real time during exercise and count some common exercises. This paper realizes real-time ROI(Region of Interest) detection and tracking through human pose estimation and hand key points detection model, which is used to calculate heart rate and blood pressure, and uses the estimated maximum heart rate to give the limit of exercise intensity. In addition, human posture estimation is also used to count movements such as squats and exercise bikes. Compared with the heart rate measured by the smart bracelet, the error is 3.4% ~ 6.7%. Under static, compared with the Omron electronic sphygmomanometer, the systolic blood pressure error is -7.39% ~ 3.43%, and the diastolic blood pressure is -6.67 ~ 4.53%. This system has low equipment requirements and low environmental requirements. It is easy to operate and does not cause any discomfort. It is suitable for daily life and will not interfere with normal activities.