Give This MIT Algorithm a Picture of Food And It’ll Give You the Recipe

Predicting recipes from photos.


MIT’s computer Science and Artificial Intelligence Laboratory (CSAIL) have Developed an algorithm, which can find a recipe based on nothing more than a single picture of food.

It is like you have seen (or eaten) a delicious meal in a picture of food and wished you had the recipe to make it.

Scientists built this algorithm by combing over a million different recipes collected from the various recipes. They then intermingle this recipe encyclopedia with image recognition software. The software identifies the type of food in a photo and matches it with its analogous recipe.

While testing, researchers found that the algorithm could correctly identify the correct recipe with 65% precision of the time. And it was most successful with pastries and other baked products as their recipes in the database and ingredients were easiest to identify.

Give This MIT Algorithm a Picture of Food And It’ll Give You the Recipe
Pic2Recipe predicts recipes from photos using a neural network, a way to achieve machine learning in which a computer learns to perform some task by analyzing examples.

Now scientists using the algorithm to identify complex dishes like sushi rolls or dishes with ambiguous ingredients like smoothies.

Researchers are now trying to improve the algorithm for great accuracy. Still, they believe that 65% success rate isn’t bad for a prototype. Although further refinement will improve the accuracy even better.

Now, scientists are planning to provide the option for users to specify some of the ingredients included in the recipe.

Lead study author Nick Hynes said, “In future, we could use this algorithm, dubbed “Pic2Recipe,” to find a recipe for all type of meals.”

“If you know what ingredients went into a dish but not the amount you can take a photo, enter the ingredients, and run the model to find a similar recipe with known quantities, and then use that information to approximate your own meal.”