Researchers at Carnegie Mellon University (CMU) have developed a method for tracking the locations of multiple individuals within an indoor setting that has implications not only for healthcare, but broader security also.
Using a network of video cameras, CMU researchers were able to follow the movements of 13 people within a senior care facility, even though individuals sometimes slipped out of camera view.
Researchers used multiple cues to identify individuals from the video feed, such as apparel color, person detection, movement trajectory and even facial recognition.
Multi-camera, multi-object tracking has been an active field of research for a decade, but automated techniques have only focused on well-controlled lab environments.
Carnegie Mellon’s team, by contrast, proved their technique with actual residents and employees in a nursing facility—with camera views compromised by long hallways, doorways, people mingling in the hallways, variations in lighting and too few cameras to provide comprehensive, overlapping views.
The algorithm developed by CMU significantly improved on two leading algorithms in multi-camera, multi-object tracking. It located individuals within one meter of their actual position 88% of the time, compared with 35% and 56% for the other algorithms.
The technology has drawn comparisons to the fictional Marauder’s Map used by popular book and movie series Harry Potter, which upon viewing and with a hint of magic, reveals the location of people moving about nearby.
While no magic was necessary to create the video monitoring tech, researches developed their tracking technique as part of an effort to monitor the health of individuals in senior care settings.
“We thought it would be easy, but it turned out to be incredibly challenging,” said Alexander Hauptmann, principal systems scientist in the Computer Sciences Department (CSD) of CMU.
Something as simple as tracking based on color of clothing proved difficult, for instance, because the same color apparel can appear different to cameras in different locations, depending on variations in lighting. Additionally, a camera’s view of an individuals can often be blocked by other people passing in hallways or even by a piece of furniture.
This is where face detection helped in re-identifying individuals on different cameras.
Yi Yang, a CSD post-doctoral researcher, noticed that faces can be recognized in less than 10% of the video frames. So, researchers developed mathematical models that enabled them to combine information, such as appearance, facial recognition and motion trajectories.
Further work will be necessary to extend the technique during longer periods of time and enable real-time monitoring, CMU researchers agree, especially as they look for additional ways to use video to monitor resident activity while preserving privacy.
“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,” Hauptmann said.
Written by Jason Oliva