These days, who hasn’t taken a “foodstagram”? Whether it’s a home-cooked meal or a particularly photogenic spread of sushi at a local Japanese restaurant, we’re all guilty of snapping away and uploading to social media. But would you be more conscious of your digital food photo exploits if you knew that the Internet could count the calories just by looking at your picture? That’s the idea behind Im2Calories, the new initiative announced by Google’s research team at this week’s Re-Work Deep Learning Summit, an annual technology conference.Although the project is still being developed, the team is working on a new advancement in artificial intelligence that would be able to count calories in your food photos by analyzing the pixels.
The Google team explains that this new technology, which “marries visual analysis with pattern recognition,” is not intended to shame food photo addicts, but to make it easier to create a digital food diary, according to Popular Science. Right now, the software only works on low-resolution photos like Instagram uploads, but the team is developing a prototype that would work for any photo. As it stands, the technology is not perfect.
“We semi-automate,” Google research scientist Kevin Murphy said during the presentation of Im2Calories, noting that you can correct the software if it misreads a photo. “If it only works 30 percent of the time, it's enough that people will start using it, we'll collect data, and it'll get better over time.”