Monitoring Hospital Rooms for Safety Using Depth Images
A two-stage fall detection technique developed by our team was tested in a real hospital setting with falls acted out in a patient room. To further test the algorithm, data were collected at the University of Missouri hospital with actual patients. Features extracted from three dimensional point clouds created from Kinect depth images were used as input to the fall detection system. Kinect sensors were placed in six hospital rooms and data were collected. The data processed from the hospital setting are discussed, demonstrating the need for an automated fall detection system which has shown robustness in addressing real world challenges in a dynamic environment.
Rantz, M. J.,
& Scott, S.
(2012). Monitoring Hospital Rooms for Safety Using Depth Images. Proceedings of the AAAI Fall Symposium Series - AI for Gerontechnology.