The app analyzes the shape and color of what's on your plate, then queries a database of 100,000 different foods to derive a calorie count.

There's an app for fat. Dieting apps are nothing new—one of the more popular categories, in fact—but most require their users to enter a lot of detailed information. But a forthcoming dieting app from Japan's NTT Communications makes counting calories a snap by deriving calorie counts from a simple photograph of the food you're eating. The app is being tested now, and is expected to be available publicly in a free beta version in January.

The app analyzes the shape and color of what's on your plate, then queries a database of 100,000 different foods to determine how many calories you're eating. It will also size up your portion and prorate your calorie count depending on what it sees, suggests Reuters. Think of it as a Shazam for food.

The report raises as many questions as answers, for now. Object recognition is notoriously difficult, even for regularly shaped objects. Food changes shape depending on how it's prepared, and depending on what angle it's viewed from. A bit of food dye—or a nice beet marinade—is sufficient to alter a dish's color. How will the app take all this into account?

An unnamed spokeswoman suggested that food the app is most familiar with is easier to recognize: "It's really good on things that Japanese people eat a lot of, like ramen, but not so good on stuff like Thai food," she told Reuters.

The app also intends to integrate social networks, since there's nothing like peer pressure to motivate. "People can see how many calories their friends have eaten, which may make them try harder," the spokeswoman also said.

The app should capitalize on a growing Facebook trend: the "look-at-what's-on-my-plate" photograph. Maybe the best way for the app-makers to grow their database of food would be to analyze the growing amount of amateur food pornography proliferating now on the web.


Republished with permission from FastCompany.com. Authored by David Zax. Photo via ccaviness/Flickr.