Software uses tiny cameras to track who’s watching digital ad displays
Digital advertising is ubiquitous these days: You can watch TV on the subway platform, in the shopping mall and even in some elevators. And while you might not be aware of it, those TV screens are often watching you.
A company called CognoVision has developed software that analyzes the view seen by a tiny camera next to the screen. Developed by a U of T computer science grad, the software can tell how many faces are watching the screen, and for how long. It can also determine, with about 85 per cent accuracy, if the viewer is male or female, and can estimate their age – vital information for advertisers trying to reach a specific demographic.
The program can also change the ads on the fly in response to the audience, explains Shahzad Malik (PhD 2007), one of the three founders of the Markham, Ontario–based company, and the brains behind the software. “For example, if there are mostly women in front of a display, the software might choose an ad for cosmetics, while a male-dominant crowd could trigger an ad for a sports car,” he says.
CognoVision has also developed an overhead tracking system that uses ceiling-mounted cameras to monitor customer flow through retail spaces such as malls and grocery stores. The software can track the number of people that enter a store each hour, which parts of the store they go to and how long they have to wait at the checkout – information that owners can use to better organize their retail spaces.
Founded in 2006, CognoVision now counts the Canadian chains Whole Foods and Pizza Pizza among its clients. After significant startup costs, Malik says he expects the company to start showing a profit in 2010.
The technology may evoke Big Brother, but Malik stresses that the people the camera “sees” remain completely anonymous. “It doesn’t know whose face it’s looking at,” he says. “All it knows is, ‘that’s a face.’”
Watch a demonstration of how CognoVision’s system works: