Hand disinfection detection using 2D image footage.

Inside medical environments, hand hygiene is very important to avoid healthcare-associated infections. The traditional method to measure hand sanitization guidelines consists of having an external observer watching the medical personnel, which can introduce bias and can only make the observations for a fixed amount of time. Some alternatives have been proposed, using different sensors and machine learning to replace the role of the observer, placing many sensors, or assuming that hand hygiene occurs in a fixed place, like a gel dispenser. In this paper, we report on our approach to detection and tracking of medical staff, as well as measuring their hand disinfection compliance using minimal equipment, obtaining only 2D images from one camera to avoid cluttering hospital space or changing the personnel’s regular behavior. The procedure is based on detecting a person in each frame, analyzing different zones with different sizes inside the same image to optimize the detection and extract as much data as possible; then tracking each recognized person and finally recognizing the hand sanitization by the motion pattern of the wrists, achieving promising results for this task in this setup.