Extraction of Recipes from Food Images by Using CNN Algorithm

Food is a primary commodity for the livelihood of mankind. The routine of humankind pursuits like food consumption, cooking, and presenting recipes on cookery show plays a major role in our daily lives. Where there are over a decillion diverse cuisine which furcates to create hundreds of niche styles of diversified flavors, tastes, and appearance. People appreciate food and enjoy food photography. So, they post food images on social media platforms like Instagram. Unfortunately, looking solely at the food image, it is not possible to comprehend the approach to its preparation technique. This paper explains about a Reverse Cooking system that revives the cooking recipes in the form of recipe names, ingredients, and cooking procedures from the input food image. The entire model is evaluated on a large-scale diverse recipe dataset and generates highly accurate predictions by leveraging the images of food. A key challenge may pose during this model building as dishes of distinct cuisines may look similar regarding their cooking procedures. To solve this, the model has been trained using Convolution Neural Networks to categorize the food images into various categories and output a matched recipe.