MIT’s New Neural Network is Shazam for Recipes

Because smart kids want to cook, too

By Ian Burke

Published on July 27, 2017

Just months after inventing deliciously scientific and deeply satisfying shapeshifting pasta, the folks at the Massachusetts Institute of Technology have come up with a new food-centric invention—a neural network that turns pictures of food into recipes.

As reported by The Verge, researchers from MIT's Computer Science and Artificial Intelligence Laboratory (CSAIL) have created a platform called Pic2Recipe that scans a picture of food and then scours the web in search of top-performing recipes based on that photo.

This seemed like it could be great for deciphering the ingredients in an ambiguous goulash or suspicious souffle, butt after walking around the test kitchen and snapping a few photos, it seems like it might be a little while before it's a usable tool. The app was able to recognize a cheeseburger, but for pork sugo (a pasta sauce), Pic2Recipe suggested a vegetable chili, and spinach ravioli in a white wine sauce returned no matches.

Now, this is clearly a prototype, and the ability to produce a cheeseburger recipe from just a picture was impressive. We’ll keep an eye out for future updates.

Want more SAVEUR?

Get our favorite recipes, stories, and more delivered to your inbox.