Carnegie Mellon University research teams have devised a method to accurately track locations of multiple people in complicated indoor settings using a complex network of cameras.
The newly devised method is able to track the movement of 13 individuals in a nursing home setting, even in situations were someone slipped out of the camera’s line of sight. Researchers utilized cues from the feed, most importantly facial recognition, to track multiple people.
Despite the fact that multi-camera/object tracking is a far from new field, until now, automated recorded techniques have only been applied to controlled lab settings. The Carnegie Mellon research team, however, applied and demonstrated their techniques with residents and employees if a nursing facility, a setting where camera line of sight is obstructed by doorways, narrow hallways, poor lighting, and overlapping views.
The Carnegie Mellon algorithm improved on the two previous leading algorithms. Carnegie Mellon’s technique located individuals within just one meter of their accurate positions 88% of the time, compared to 35% and 56% for the previous algorithms.
Researchers, Alexander Hauptmann, Shoou-I Yu, and Yi Yang will present the team’s findings at June 27th at the Computer Vision and Pattern Recognition Conference in Portland Oregon.
"The goal is not to be Big Brother, but to alert the caregivers of subtle changes in activity levels or behaviors that indicate a change of health status," Alexander Hauptmann commented. All of the people in this study consented to being tracked.
While potentially controversial, this system could revolutionize in-home or nursing home care.