Database for facial video-based Respiratory Rate Recognition
The database was created to enable measurements of respiratory rate from facial video images. It includes recordings of 12 adults ages ranging from 23 to 36 years. There are 10 men and 2 women. Four videos per person were recorded with an RGB camera (Pike F-145) from a distance of about 1.5 meters for 3 minutes per sequence with a frame rate of 25 frames per second. Four different respiratory rate scenarios were recorded for every subject. In the first scenario the spontaneous respiration of the test person is recorded. For the next three measurements, the subjects should try to follow a given respiratory pattern, which was displayed on a monitor in front of them. The respiratory frequencies for the single videos were 10, 15 and 20 breaths per minute. Over all measurements the participants were asked not to move heavily. The reference signals were recorded via a chest belt (NeXus NX-RSP1A) with a sample rate of 512 Hz using a trigger signal for synchronization with the camera.
Publications
Marc-André Fiedler, Michal Rapczynski, Ayoub Al-Hamadi
Fusion-Based Approach for Respiratory Rate Recognition From Facial Video Images
IEEE Access, 2020.