The following is an excerpt from the original THE WALL STREET JOURNAL article.
...
Already, video images can be subpoenaed from retailers for law-enforcement purposes. Technology capable of matching a photo with an individual's identity, say from credit-card transactions, "has certainly arrived," says Rajeev Sharma, a Penn State University computer science professor who has launched a company that is creating shopper-monitoring systems. It isn't certain whether retailers are availing themselves of the know-how. Credit card companies currently aren't sharing individuals' financial information with retailers, he adds, but retailers have their own customer databases as the result of loyalty cards, store credit cards and other in-house programs. Theoretically, they could link a transaction at a cash register with the face of a shopper appearing on the videotape.
Dr. Sharma's start-up, Advanced Interfaces Inc., of State College, Pa., is expected this week to launch a Web site, videomining.com, highlighting the company's patented "computer vision" technologies.
In a pilot project conducted last year in the Philadelphia area, Advanced Interfaces set up nine cameras in each of two McDonald's Corp. restaurants to find out which consumer types would find a new salad item most appealing. The research was done without consumers' knowledge, says Dr. Sharma, who is Advanced Interfaces' chief executive.
Seven of the cameras were already in place for security purposes and needed only to be reconfigured using Advanced's sensors. Two additional cameras were positioned in the ceiling directly over cash registers. By measuring the shapes of people's faces, the sensors were able to provide a breakdown of the fast-food customers by race, gender and age group, he says. The videos also revealed the length of time customers spent waiting in line or looking at the menu before ordering. Mr. Sharma declined to discuss the findings.
All of the video was subsequently destroyed, he says. "Only the computers and no humans saw the pictures of the customers," Mr. Sharma says. Advanced is conducting similar consumer-behavior analysis this holiday season for three other retailers that Mr. Sharma declined to identify.
Video mining is being spurred by digital video cameras. Unlike their analog counterparts, digital video cameras can be programmed so that the images can be quickly read by computers -- taking only hours to complete tasks that might have taken weeks for humans to do.
...
