Eye-tracking device watches you watching ads

A Computing & Information Science Professor from Queen's University in Ontario, Canada, has developed an-eye tracking device that is set to revolutionise current billboard and screen advertising sales models, and has the potential to pave the way for a future where personalised ads are pitched direct to individuals.

The portable device, dubbed the eyebox2, can be attached to public area advertisements and uses a camera that monitors eye movements in real time to automatically detect when people are looking at it from up to 10 meters away and at a horizontal range of 2-3 meters.

By emitting infrared diodes and recognising the red-eye effect, the device mimics eye contact perception in humans, allowing it to accurately pinpoint what television screen, billboard or product shelf people are looking at.

This enables advertisers to track the number of people who engage eye contact with their ads said Professor Roel Vertegaal, the chief developer of the eyebox2 and director of the Human Media Laboratory at Queen's University.

"The information is filtered such that only fixations are counted. We can maintain statistics on the number of people as well as the average time per person," said Vertegaal. "We think this will revolutionise the ambient advertising business in a way that web analytics have revolutionised the online ad business."

While Internet ad success can be measured by the number of hits on a web site, it is much harder to assess the effectiveness of television screens that target people in shopping centres, movie theatres and other public places. But the eyebox2 is set to change this by giving advertisers a tool to accurately measure how much attention an advertisement or product receives, not in the number of hits, but in the number of eyeballs.

Vertegaal said this would enable brick-and-mortar stores to use an advertising revenue model similar to Google's online PageRank and web analytics technologies, opening up the possibility for advertisers to now be charged on a "per look" basis.

He said the eyebox2 would also work well in home automation and security environments. However, variations of the latter application may prove to be the contentious issue considering the lack of standards in place to govern the eyebox2's potential ability to identity individuals.

In one possible scenario, it's easy to imagine a device like this being used to not only identify individuals but also pitch personalised ads. Such a scenario conjures scenes from the futuristic crime thriller film, Minority Report where, Tom Cruise's character, who is on the run from the law, must undergo eye-transplant surgery to avoid being identified by advertisements that are pitched direct to individuals based on retinal scans. Although it may seem a far away possibility, the Big Brother undertones and intrusive nature of the technology cannot be overlooked.

However, Vertegaal argued that Orwellian uses of the eyebox2 were not possible, as the device's resolution is not sufficient enough to capture retinal scans and in any case, was never created for such uses.

"We specifically steered clear from this well-known [identity tracking] scenario. eyebox2 does not and will not do iris or retinal scanning," he said. "As such, it is as harmless as a door-sensor because the resolution is insufficient for this functionality."

Xuuk, the company in charge of commercialising the technology, is doing its best to avoid such scenarios when dealing with the university and advertising agencies it has already presold the device to.

"We clearly set boundaries with large accounts we deal with, for example, [in its present form] it is not possible to capture images with our device," said Vertegaal, who is also Xuuk's CEO.

But although Vertegaal ruled out the marriage of the eyebox2 technology with retina scanners or image capturing devices, he conceded the possibility was out there and warned that if customers chose to combine the eyebox2 technology with other image capturing devices, there was little his company could do about it.

"[Already], face recognition software is being used in Europe to track shopping mall theft," he said. "While we do not encourage such use, and given that our cameras cannot identify people or provide images, it still seems these directions are already being taken by other companies regardless of our hardware."