DIET ALCHEMIST

Abstract

Food is an essential aspect of human existence, and individuals are constantly exploring new and delectable dishes. Often, people choose food items from grocery stores without knowing their names or recognizing them immediately. Therefore, it is crucial to comprehend the elements that can be combined to create delightful recipes. Selecting the appropriate recipe from alist of ingredients poses a significant challenge for both novice and expert chefs. Machine learning plays a prominent role in our daily lives, such as in object recognition through image processing. However, traditional methods employed in this process, which involve numerous food items, present a higher risk of error. To address these challenges, we developed a model that identifies food ingredients and formulated an algorithm to recommend recipes based on the identified ingredients. Our research involved constructing a unique dataset comprising 9,856 photos, categorized into 32 types of food items. We utilized a Convolutional Neural Network (CNN) model for food item recognition and employed machine learning techniques to generate recipes. Our approach achieved an impressive accuracy rate of 94 percent, which holds significant practical value. Keywords: food ingredients, recipe recommendation, machine learning, deep learning, Convolutional Neural Network (CNN), object recognition, image processing.