UW team develops camera technology that can track a person from one camera to the next
SEATTLE — After the Boston Marathon bombing in 2013, investigators sifted through hours of video from nearby surveillance cameras trying to find their suspects.
Now, imagine a technology that would cut the hunt down to a matter of hours or even minutes.
University of Washington Electrical Engineering professor Jeng-Neng Hwang has developed a technology to make that happen.
“We can quickly identify this person has been coming from here, put the backpack here, walking away to where,” Hwang said. “So this can be a wider area search — everything by computer instead of police or investigators.”
The technology allows unrelated cameras to talk to each other. He can now track individuals on different cameras — following a person’s trajectory from one camera to the next.
“So when a person being tracked inside one camera, this person gets out of this camera, this person may walk to any other camera. When the person enters into whichever camera, that camera immediately knows the difference between the original camera and this new camera,” Hwang said.
Let's say the person you're looking for is in a blue coat, Hwang's program allows each camera to detect that color, and the distinctive movement of a particular person.
"So they can transform whatever useful information into that other camera's view and it immediately says, yeah, that's the same person," he said.
Hwang also wants to pair the technology with a system like Google Earth, creating a virtual 3-D map in real time. It could even help police track a suspect on the run.