FOOD INGREDIENTS PREDICTION AND RECIPE GENERATION FROM IMAGES

Abstract

Food is the basic need for all living beings to survive. Cooking is a fundamental task in a day-to-day activity of human beings. There are many traditional foods whose ingredients and the recipe have lost. People who like cooking are always searching for new food recipes to cook. Many people are into food photography. We can find a lot of images of food on the internet and social media which we want to cook but cannot find its recipe. It is not easy for people to determine the food ingredients and cooking instructions by just looking at the image. To address this issue we present ”Food Ingredients Prediction and Recipe Generation system”. With the help of SOTA Computer Vision and Natural Language Processing methods, this system uses a large-scale dataset ”Recipe1M” of food ingredients, instructions, and images for establishing a relationship between them and predict the food ingredients and generate recipes for the input images.