Google has unveiled a new project, called Im2Calories, which will use deep learning algorithms to estimate how many calories are in food, based on a photo of the meal, according to a report from Popular Science.
The algorithm will estimate calories in a meal, which could be, for example, eggs, pancakes, and bacon, based on the size of the individual items compared to the size of the plate. Pictures do not have to be high resolution for the system to work either -- a shot from a smartphone would work as well.
Google Research Scientist Kevin Murphy, who discussed this project at an event in Boston last week explained that the algorithm is not very accurate, but the more pictures people take, the more it will improve.
“To me it's obvious that people really want this and this is really useful,” he said at the event. “Ok fine, maybe we get the calories off by 20 percent. It doesn't matter. We're going to average over a week or a month or a year. And now we can start to potentially join information from multiple people and start to do population level statistics. I have colleagues in epidemiology and public health, and they really want this stuff.”
Murphy added that in the future they want to use this technology to do scene analysis and offer users a traffic scene analysis and a prediction of where the most likely parking spot is.
Other companies have incorporated photo taking into nutrition tracking offerings in recent years. A similar tool, called Meal Snap, which MobiHealthNews included in a roundup last year, aimed to calculate what’s in food and how many calories it has, based on food photos. This app is no longer available in the US Apple App Store.
Another app, called Eat It Tweet It, integrates with Twitter and offers users a portal to tweet exclusively about their diet. The app includes preset hashtags and a camera feature to take a picture of the meal. A study of the app showed that young adults found it useful for tracking food consumption, though that study didn’t look at weight loss outcomes.
And The Eatery, an app developed by Massive Health, which Jawbone acquired in February 2013, also used a smartphone camera as the input mechanism, but rather than calculating the nutritional value with an algorithm, it crowdsourced the food rating from other Eatery users who rated how healthy the meals were on a sliding scale.