‘Shazam for Photos’ IDs Famous Faces for Paparazzi

Steve Eichner got tired of answering the same question over and over again. As a veteran photographer for the fashion biz’s Women’s Wear Daily, there was a constant tap-tap-tapping on his shoulder as he photographed models and socialites. The solution was his new service, NameFace.com, which has already been called the “Shazam for photography” for its ability to answer the question, “I know that guy, who is that?”

“I’d be at a fashion show, and the younger photographers would tap me on the shoulder and say, ‘Whozzat? Whozzat? Whozzat?’” Eichner said. “After a while I’d say, ‘I’ll give you the first name for free, and if you want the last, you’ve got to pay me a dollar.’ But it was time to solve that problem for good. I can’t shoot pictures if I’m stopping to play name-tag with everyone.”

That was the inspiration. The perspiration came from Daniela Kirsch, a German coder helming Hello Plugin, a web-dev company. The two met via Tinder while she was on vacation; when they paused their merrymaking to chat, she found herself entranced with the idea, and coded the beta version after her return to Berlin.

This week, the two launched their collaboration, a facial recognition service based on software she built from the ground up.

“There are companies out there that do facial recognition,” Kirsch said, “but for government and security — not for the entertainment industry. Most are based on what basically looks like a mug shot, but trying to recognize someone ‘in the wild’ [who’s] moving and dancing and the head is at different angles, that is really new.”

Another difference between this and other facial recognition is the privacy factor. “Google Glass was doing facial recognition to identify private citizens and bring up their Facebook profiles and whatnot,” Eichner added. “We only identify public figures who are out at parties to be seen. Nobody has ever applied facial recognition to this use case: To identify public figures for event photographers.”

The target user is a paparazzo who works red-carpet events and wants to upload and distribute photos to media outlets or to services like Getty as quickly and accurately as possible. The ideal use case would be, say, a huge Fashion Week party glittering with luminaries of various strengths — particularly those on lists B through D, whose names might not leap to mind easily.

Right now, Google Images can only identify photos based on composition and color, not actual faces. The Facebook tagging feature works with a select group of connections with limited success. The potential of the technology has barely been tapped, says Kirsch.

“We can take the guest list of an upcoming party — not just the models, but the company executives and VIPs — and, with human help, match names to faces ahead of time,” Eichner said, adding that the company has a small team of coders and six employees dedicated mostly to training the database.

Right now, the service is free; eventually, it will become a tiered subscription model, depending on how quickly a photographer wants to upload, how much customer service they want and other features. The service should be useful not just to photographers, but public relations agencies, photo databases and entertainment magazines and websites.

Eichner’s roots in online photography services run deep. In 1995, he and two friends from Columbia University registered Photography.com to launch the first online stock photography agency — an idea that turned out to be ahead of its time, as the magazines he worked with still preferred to work with slides and prints. He ended up leasing the name out for a number of years, then selling it in 2006. (Full disclosure: I wrote the website and marketing copy for the first iteration of Photography.com.)

One more fun fact about Steve Eichner: His baby half-brother, via his father, is Billy Eichner of Billy on the Street. “I used to give him Led Zeppelin CDs, and he’d tell me ‘no, go back and get me Madonna,’” said Steve. So yeah. That.

This article originally appeared on Recode.net.