Making (and Measuring) Eye Contact

LOS ANGELES Steven Spielberg’s depiction of mall signs that identified potential customers by retinal scans in Minority Report apparently wasn’t far off. Scientists at a Queen’s University, Toronto, have developed a technology for determining whether signage is catching the eyes of passersby—or not.

Called Eyebox2, the technology uses infrared-scanning light-emitting diodes to pick up the “red eye” effect, plus a pattern-recognition algorithm to measure the number of people that pass by the sign.

Next month, the first version of “eyeanalytics” software rolls out, said Dr. Roel Vertegaal, president and CEO of Xuul Inc., who works at the Human Media Lab on the Kingston campus. The software running on the university’s server collates metrix from the boards and allows agencies to follow the progress of campaigns over the Internet, he said.

“We’ve sold quite a bit of hardware, but this has been the missing link,” Vertegaal said. “[Users] can get all the key statistics, including the average time people spent in front of the sign. They can also filter the information by campaign or by sign location.”

Vertegaal said the system is greater than 80 percent accurate at measuring eye contact within 10 meters (33 feet) and 96 percent accurate for counting passing faces.

He said the ability to see whether display advertising is working would allow agencies to adjust campaigns accordingly, based on the “look of the screen” with respect to the behavior of passersby it inspires.

Among agencies involved in testing the system are Novramedia in Toronto, Vertegaal said.