This AI Can Tell The Ingredients Of A Dish By Looking At A Photo

Love taking photos of your food? Soon we'll be getting more than just a photo.


Nowadays, most people just love taking photos of their food and posting it on social media. Usually, it’s just for the likes and nothing more. Well, researchers from MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL) and Qatar Computing Research Institute (QCRI) took it to the next level by training an artificial intelligence system to look at a photo of food and predict the ingredients as well as suggest similar recipes.

The researchers call this system Pic2Recipe and they believe that by analyzing photos of food, we could learn recipes and understand people’s eating habits.

According to Yusuf Aytar, an MIT postdoc who co-wrote a paper about this system with MIT Professor Antonio Torralba. “In computer vision, food is mostly neglected because we don’t have the large-scale datasets needed to make predictions. But seemingly useless photos on social media can actually provide valuable insight into health habits and dietary preferences.”

Source: YouTube, CSAIL

How it works

The team combined websites like All Recipes and Food.com to create a database of over 1 million recipes, “Recipe1M”. The team then used this data to train a neural network to find patterns and create connections between the food images and their corresponding ingredients and recipes.

When Pic2Recipe is given a food item, it is able identify ingredients like butter, eggs, floor etc, and will suggest several recipes that the AI thinks is similar to images in its database. The system did good with desserts like cookies and muffins, since this was a main theme in the database. On the other hand, it had some difficulty determining ingredients for more ambiguous foods smoothies.

Source: Kidspot

The team hopes to improve the system in the near future so that it will be able to understand food further. From knowing how a certain food is prepared like boiled or fried, or distinguish different variation of foods. The researchers also like the idea of developing the system into a “dinner aide” that can help the user figure out what to cook when it is given a dietary preference or maybe given a list of items in the fridge.

This can be very handy for people who want to figure out what’s in their food when they don’t know a certain food’s nutritional information.

 

Read more  Sticky Labels on Fresh Produce Are Getting Replaced With Laser Marks

Article Source:

MIT CSAIL

Comments 0

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.

This AI Can Tell The Ingredients Of A Dish By Looking At A Photo

Send this to a friend