At the Rework Deep Learning Summit in Boston, Google’s Kevin Murphy revealed a project called Im2Calories that will analyze a photograph of food and report how many calories are on the plate.
Im2Calories, which can use any relatively clear smartphone photo, aims to make keeping a food diary easier and more accurate, and hopefully combat the obesity epidemic by encouraging people to pay closer attention to what they’re eating and how much. The program will measure food against the size of the plate to determine serving sizes. If its guess is inaccurate, users can correct it with a drop-down menu, says Hannah Bouckley of BT.
In 2014, Google acquired UK-based DeepMind for $ 400 million, presumably to assimilate their experience with deep learning, writes Erik Sofge of Popular Science. Deep learning programs can improve themselves as they are used and corrected, and Im2Calories is a prime example.
To me it’s obvious that people really want this and this is really useful. Ok fine, maybe we get the calories off by 20%. 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.
Deep learning paired with visual analysis could grow to become an extremely valuable tool, with implications for artificial intelligence as a whole. SRI International, who developed Apple’s Siri, announced last year that they were working to produce a similar calorie-counting app, and developers are looking forward to applying this technology to a wide range of other uses.
If we can do this for food, that’s just the killer app. Suppose we did street scene analysis. We don’t want to just say there are cars in this intersection. That’s boring. We want to do things like localize cars, count the cars, get attributes of the cars, which way they are facing. Then we can do things like traffic scene analysis, predict where the most likely parking spot is. And since this is all learned from data, the technology is the same, you just change the data.
However, the visual nature of the app means that developers are having trouble accurately assessing invisible calories, like butter and oil, according to Ellie Zolfagharifard for the Daily Mail. This is one of the hurdles that developers must conquer before Im2Calories is released.
It has not yet been announced when this app will be available to the public, writes Rex Macadangdang of Tech Times.