Facing a fridge full of ingredients but still don’t know what to cook? Tired of following the same recipes and eager to try something new and creative? Thanks to AI technologies such as image recognition and machine learning, people can now save time, food and money in the kitchen while discovering creative and tasty recipes and even generating their own new and personalized flavours.
Looks good, I’ll try: Generating recipes from photos
Facebook has developed an image-to-recipe generation system which enables users to reverse engineer a recipe by simply inputting an image of the dish they want to prepare. First, ingredients and ingredient co-occurrence are generated by exploiting visual features extracted from the food image. The system generates an ingredients list and preparation instructions based on the predicted ingredients and the image’s visual features. Evaluated on the large-scale Recipe1M dataset, the system showed good performance compared to previous baselines for ingredient prediction.
Pot Luck: Making a meal using only the ingredients in your fridge
Combining AI technologies along with gastronomical learning, companies like Plant Jammer and Chefling help people create tasty and personalized recipes using ingredients they already have on hand. Based on algorithms built on databases with flavor tags, ingredient processing methods and recipe steps, the practical system recommends recipes based on available ingredients and the user’s personal preferences and requirements re flavours, nutritional values, and cooking methods. Users can interface via personal assistants such as Google Assistant or Amazon’s Alexa. The system integrates with smart home appliances such as refrigerators and ovens to track available ingredients and cooking temperatures and times — transforming almost anyone into a versatile home chef.
Creating new and improved flavors through science
IBM Research and McCormick & Company have created a novel AI system to help developers create new flavours more efficiently and effectively. Designing new flavour experiences is challenging since how humans experience flavours is a topic that needs much further exploration. It is however now possible to use AI to effectively pair flavours based on the wealth of available data accumulated over decades of food operations, including for example historical flavor formulas, raw material components, experiment and consumer test results, market success and more.
The system designed by IBM and McCormick mainly uses advanced machine learning algorithms to learn and predict raw materials and formula combinations, helping product developers optimize current flavor formulas or develop new flavor formulas tailored to specific preferences. The products developed with the help of the “McCormick ONE” AI system are already available in retail and have been favourably received by customers, the company says.
While deep learning and smart home technologies can save time and reduce waste in the kitchen, such integrations could also reduce the enjoyment traditionally associated with preparing food with families and friends. Like playing a piano or reciting a poem, cooking is a personal expression and will likely remain so. It’s also impossible to prepare a dish that will be loved by everyone, food preference is very much a matter of personal taste. So while AI technologies can certainly provide advice and inspiration in the kitchen, it’s expected humans will still wear the big chef’s hat for the foreseeable future.
Author: Kelly Xie | Editor: Michael Sarazen