Google technology counts calories on a plate by analysing a photo

(File Photo): Photographing their food in a café with their smartphones.
(File Photo): Customers photographing their food in a café with their smartphones.

Unsure whether the food you are eating is laden with calories? Google is using artificial intelligence to tell from a simple photograph.

BOSTON – The rise in popularity of mobile phones has bred several photo trends: retro filters, selfies and food photography, notably in photo-sharing app Instagram.

Google is taking the food snap to the next level by developing software which can determine how many calories are in your meal simply by analysing a photograph of it.

The project, called Im2Calories, is being run by researchers within Google’s artificial intelligence (AI) division and was revealed at the Rework Deep Learning Summit by research scientist Kevin Murphy.

According to Popular Science, the software works using image-recognition technology that analyses the depth of each pixel, matching it with patterns it recognises and existing calorie information.

Food items aren’t always the same size, of course (for example, a burger varies in size and thickness), so the software analyses the size in relation to the plate and any condiments.

It’s not totally automatic, but if the software incorrectly recognises a piece of food, you can correct it using a drop-down menu.

Murphy admitted Im2Calories isn’t 100% accurate: “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.”

As well as being a useful tool for dieters, Murphy believes the information could be useful on a bigger scale.

“We can start to potentially join information from multiple people and start to do population level statistics,” he said. “I have colleagues in epidemiology and public health, and they really want this stuff.”

The technology could also be used beyond food, analysing streets.

“We want to do things like localise cars, count the cars, get attributes of the cars, [and] which way are they facing. Then we can do things like traffic scene analysis, predict where the most likely parking spot is,” said Murphy. “And since this is all learned from data, the technology is the same, you just change the data.”

Google has filed a patent for Im2Calories, but there’s no news yet when it will be available to the public.