Tell us about the project. We launched a digital outdoor campaign for Women’s Aid to tackle domestic violence. Digital screens went live across London’s Canary Wharf and Westfield and Birmingham’s Bullring, showing a woman with a battered and bruised face and a caption that said: "Look at me." The billboards were equipped with facial-recognition technology so the more people looked at them, the more the woman’s wounds healed.
How did you come up with the idea? When we looked at Ocean Outdoor’s screens and their technical spec, we found that we could use them to deduce when people were walking past and when they were looking at the screen. We had previously supported Women’s Aid on a campaign that urged viewers: "Don’t turn a blind eye." This seemed like a clear opportunity to make a difference by literally not turning a blind eye to domestic violence.
How did the technology work? Much of the technology was already installed in the billboards, with cameras generating a constant stream of data from passers-by. We created a Flash-based platform that listened to the camera’s data, pulled out the face information and triggered a series of seamless video sequences to play when enough faces had been detected. The video sequences, photographed by Rankin, created seamless loops and transitions. It then created the illusion of a live victim on screen that reacted to actions from passers-by. We also used the GPS location to send donation messages to passers-by who had opted in.
What were the challenges? There were a few custom solutions needed. The video feed from the cameras needed to be converted constantly into a format that supported Flash. We also had to add code routines to dynamically adjust the progress of the ad to take account of busy and quiet footfall periods. With three different-sized screens, a lot of tuning was needed to make sure the face detection worked at the right time.
How was the feedback? The response has been incredible. There have been 86.7 million impressions on Twitter alone. There was also a 2,500 per cent increase in people stopping to watch the ad for more than ten seconds, versus the previous average for the same poster sites.