A web app that allows you to experience AI that generates recipes from cooking images
Among the papers published on
[1812.06164] Inverse Cooking: Recipe Generation from Food Images
Generate a recipe from a cooking image AI-Qiita
The authors of the paper published in arXiv are the authors of Amaia Salvador , an expert in computer vision who works for 'Facebook AI Research', an AI research unit on Facebook. The research group developed “AI automatically generating the recipe of the food in the photograph”, which predicts the ingredients as a set, models their dependencies without imposing an order, and estimates the ingredients estimated from the image It is to generate the cooking instruction (recipe) used.
Salvador publishes the source code for 'AI that automatically generates the recipe of the food in the photo' on GitHub.
GitHub-facebookresearch / inversecooking: Recipe Generation from Food Images
The source code released on GitHub has been made into a form that relu can easily try as a web application.
AI that generates a recipe from a cooking image
It's easy to use, just click on the red frame and select the food image you want to know the recipe or drag and drop the image.
The recipe generation will take some time.
The first recipe I created was McDonald's new burger ' Bakyn Davt (Yaguchi), with a great source of oyster sauce .' AI guesses the food in the picture and gives four suggestions. Yaguchi is guessed as 'bacon cheeseburger', bacon cheeseburger using Lea & Perrins 's original Worcestershire sauce, 'bacon wrapped beef sandwich', 'cheese flavored bacon burger', and detailed recipes can be checked.
Then, I think that the correct recipe will be displayed if it is a major pizza in the same way overseas as well, and use the photograph of '
The recipe for “Tomato gravy”, “Easy tomato gravy”, “Beef stew”, and “Tomato onion gravy” was output from the “Found