[4Xin1-22] Reliability Improvement Method for Remote Photoplethysmography from Face Video
Keywords:Remote photoplethysmography, Pulse rate, Machine learning
In recent years, the demand for telemedicine has led to the development of contact-less sensing technology using cameras and sensors. Remote photo plethysmography (rPPG) has been reported to be highly accurate.
The purpose of this study is to realize highly accurate rPPG in a real environment by preventing errors of rPPG. Specifically, the reliability of rPPG is judged based on images.
In this study, we collected face images in which rPPG was performed accurately and inaccurately. We built a model to judge the reliability of estimates using machine learning. RGB pixel values in the face region and the results of frequency analysis were used as features. Experimental results showed that the constructed model can judge the reliability of estimates and prevent errors. In the future, the effectiveness of this technology will be verified in a real environment.
The purpose of this study is to realize highly accurate rPPG in a real environment by preventing errors of rPPG. Specifically, the reliability of rPPG is judged based on images.
In this study, we collected face images in which rPPG was performed accurately and inaccurately. We built a model to judge the reliability of estimates using machine learning. RGB pixel values in the face region and the results of frequency analysis were used as features. Experimental results showed that the constructed model can judge the reliability of estimates and prevent errors. In the future, the effectiveness of this technology will be verified in a real environment.
Authentication for paper PDF access
A password is required to view paper PDFs. If you are a registered participant, please log on the site from Participant Log In.
You could view the PDF with entering the PDF viewing password bellow.