Camera-based Vital Signs Measurement: From its Origins to the Deep Learning Era

Simple RGB video of the human face gives away physiological information through tiny changes in skin color and motion — imperceptible to the human eye, but measurable by the computer to produce estimates of vital signs such as heart rate and respiration rate. The technology that accomplishes this is know as remote photoplethysmography (rPPG), highlighting its major advantage of not requiring physical contact between subject and sensor.

First proposed in 2008, rPPG has yet to be widely commercialized. This is in part due to the difficulty in obtaining reliable measurements in challenging real-world conditions, especially regarding illumination conditions and subject movement. This talk will (i) introduce the basis for the signals obtained with camera sensors, (ii) walk through the technological improvements made by researchers since 2008, and (iii) summarize current issues and give an outlook on future applications.