Thanks to Science, You Can Now Turn Food Porn Into Recipes

Artificial intelligence can now help you get the recipe for a dish just by uploading a photo of it.
two friends taking photo of their food with smartphones

two friends taking photo of their food with smartphones

Have you ever seen a photo of a dish someone has posted to social media and thought, “Gee, I’d like to make that” — or Instagrammed a photo of a meal at a restaurant and wondered about ingredients?

Thanks to researchers at the Massachusetts Institute of Technology’s Computer Science and Artificial Intelligence Laboratory (CSAIL) you may be able to get the recipe for a dish just by uploading a photo of it.

Using images and recipes gathered on food websites including, the researchers have compiled an expansive database of more 1 million recipes and 800,000 images. Dubbed Recipe1M, it is purported to be the largest collection of recipe data available to the public. In conjunction with a “neural network” developed by the researchers, the database can be used to provide recipes that match the visual information detected in an image.

The researchers have also created a user-friendly online version of their model — a demo of a forthcoming app called Pic2Recipe — that aims to allow people to test it out. (Alas, we couldn’t get it to work, but here’s a video that shows how it does.)

The team is still refining its food-focused artificial intelligence system, tweaking the network so it can distinguish, for instance, similar looking ingredients common to a specific dish as well as how an ingredient has been prepared (sliced vs. diced). But they foresee a range of uses for their model — from gleaning nutritional information about a meal when the ingredients are unknown to sleuthing the intel necessary to replicate a restaurant meal.

“You can imagine people using this to track their daily nutrition, or to photograph their meal at a restaurant and know what’s needed to cook it at home later,” Christoph Trattner, an assistant professor at MODUL University Vienna in the New Media Technology Department, who was not involved in the paper, said in a release. “The team’s approach works at a similar level to human judgment, which is remarkable.”

And really cool, too.

Photo courtesy of iStock
Keep Reading