Sign up for our newsletter
Transport / Aviation

New CCTV cameras can track pedestrians around cities

After two pressure cooker bombs exploded during the Boston Marathon in April 2013, killing three and injuring around 300 more, it took police five days to track down the culprits. While thousands of officers scoured a 20-block area, transport systems were shut down and residents told to stay indoors.  

Now, just over a year later, an electrical engineer from the University of Washington says he has developed technology that would have shortened that manhunt to “hours, if not minutes”. 

That engineer is Jenq-Neng Hwang, who has created a system of cameras which use algorithms to track moving objects across large areas – even when they move out of sight of one camera and into the frame of another.

Before the cameras begin tracking, they record “training footage”, to understand how objects change in angles, texture and colour when viewed from the cameras’ respective locations. Once tracking, they assign moving objects colours and numbers, and use the information from the training footage to detect people already identified by another camera.

White papers from our partners

At the moment, the cameras can’t do this in real time – the technology operates as a faster version of officers searching footage frame-by-frame for suspects. But eventually, Hwang hopes these cameras could operate around cities, on drones and in robots to track suspects or create a kind of moving Google Earth, with maps populated by cars and people. You could, for example, see the traffic on a road you’re about to travel to, or use a tracking drone to find survivors after a natural disaster.   

In the video below, Hwang admits there are “privacy issues” surrounding this vision. But to him, it seems absurd that we collect millions of hours of CCTV footage in cities, only for it to “end up on servers, never to be viewed again”.

One idea that does seem a little excessive is the use of the cameras in shops, to collect what Hwang calls “valuable information about a specific shopper’s preferences”.  While he enticingly says this information would be used to send shoppers “special coupons”, we’d probably rather shop without a robot salesman breathing down our necks.

All images: University of Washington. 
This article is from the CityMetric archive: some formatting and images may not be present.