Abstract:
Measuring the heart rate is the most essential foundation for arbitrary applications in medicine and sports, as well as in many other cases of application, like psycholog...Show MoreMetadata
Abstract:
Measuring the heart rate is the most essential foundation for arbitrary applications in medicine and sports, as well as in many other cases of application, like psychology. The conventional ways of measuring the heart rate like pulseoxymetry, electrocardiogram, wrist belts and so on are afflicted with some very own disadvantages. The most prominent point is the need of direct contact between the measurement unit and the participant, which often leads to discomfort and irritation of the skin, to mention only a few. To avoid this crucial disadvantage, several complete contact-free measurement techniques were developed in recent years. The most promising technique is the analysis of recorded video data of a humans face. While these techniques were developed further and further, they stay limited to only one participant at the same time. As a solution for this circumstance, we proposed in a multi person measurement approach using convolutional neural networks, in a previous paper. In this new paper, the first attempt is enhanced in terms of run time and accuracy by using an adapted algorithm for HR estimation. It will be presented that the evolutional approach is able to estimate the heart rate not only for one, but for a huge number of participants with a very small runtime and high accuracy.
Date of Conference: 04-06 January 2020
Date Added to IEEE Xplore: 23 March 2020
ISBN Information: