Skip to main content

A young woman in a blue suit passes a flashing billboard in central Tokyo. Within seconds, the screen flickers and an advertisement suggests a crisp glass of white wine and a plate of prosciutto at a nearby Italian restaurant.

It appear like just another ad - except there is a difference. The billboard has not only identified the passerby as a female office worker in her thirties, it's also decided she is the kind of person who would sooner choose a classy Italian restaurant over the beer-bars favoured by her male colleagues.

Welcome to the future of advertising. Lurking among the conventional advertisements that cover Tokyo are a new generation of intelligent digital billboards. Created by the electronics company NEC, they use sophisticated facial-recognition technology to tailor the ad message based on the appearance of the passerby and the surrounding environment.

And so a young women identified as a stylish 20-something may be informed about a newly opened boutique just around the corner. Or a middle-aged man walking down the street in a downpour may see a flashing image advertising the nearest store selling raincoats.

"The billboards identify consumers in terms of their age and gender as well as intelligently assessing other surrounding factors," NEC's Chris Shimizu explains, demonstrating the billboard at the company's Tokyo showroom. "Depending on whether they are wearing a suit or T-shirt and jeans, they can tell whether they are a student or a salaryman and advertise accordingly."

He added: "These billboards are a world's first in terms of the level of sophistication involved in the technology, which enables them to identify these very specific details."

The billboards bring to mind scenes once deemed farfetched from sci-fi movies such as Minority Report, in which a billboard scans Tom Cruise's iris before declaring: "John Anderton, you could use a Guinness right about now."

Tokyo's new billboards don't quite so far as to identify consumer by name, but they are likely to be just as confident in suggesting a beverage likely to entice a thirsty individual.

Key to the technology is a small camera above the screen that transforms the image of the person into data, which is then matched to a database of 10,000 profiles of real people in an inbuilt computer system.

Once matched, the technology can confirm - with 85 to 90 per cent accuracy - the person's gender and age, and take a good guess at social status and lifestyle as well.

Consumers who scan the advertisements with their mobile phones can receive restaurant menus or shopping information tailored to their identity directly in their phone.

"These systems are very flexible, depending on the client," Mr. Shimizu says. "They can be programmed to take into account the weather and time of day as well as the individual's profile."

And they are set to become increasingly common. Since the billboards were first piloted this summer, they have been bought by 10 major Japanese companies, from restaurant chains to department stores.

The company, which is also behind the facial recognition technology used by border controls between Hong Kong and China, is currently planning an entry to the North American market.

"The use of this technology in retail spaces and points of sale will become commonplace," says Tyron Giuliani, a consumer and advertising specialist with Tokyo-based consultancy Optia Partners K.K. "With leading Japanese technology, media and content partners like Softbank, Yahoo, NEC and NTT all working to roll out this technology and its applications, there will be no stopping it."

How the billboards work

1. A small camera above the main screen captures an image of the face of a passerby.

2. The distance between the consumer and the screen is determined to gauge the potential level of interest.

3. A computer transforms the facial image it into a digital data form.

4. The data are then compared to the data of 10,000 real people whose faces have previously been recorded in the computer's database, along with their lifestyle profiles.

5. The individual will be categorized by gender and one of 10 age categories.

6. The computer can also judge height and clothing style, whether they are accompanied by children, or if it is sunny or raining for further specification.

7. The computer then selects pre-programmed advertising suitable for the particular category of consumer.

Special to The Globe and Mail

Interact with The Globe