Ingredients to Recipe: A YOLO-based Object Detector and Recommendation System via Clustering Approach

Abstract

Career, school, work, new adventures of life, those are often given priority. Eating healthy becomes the next important concern, so going to a fast-food joint and preparing instant food may solve the problems related to food at the moment, but eventually it deteriorates health, either through weight fluctuation, energy loss, or both. To help overcome this, our proposed model aims to create a recipe recommendation system based on ingredient recognition. It helps explore new recipes in the kitchen for beginners, busy parents, foodies, and pro chefs alike. Our system helps users decide what they can cook with the available resources by making use of images of ingredients. YOLOv5 has been employed to detect ingredients. This enables multiple object detection in real-time. An API call is done to calculate the calorie based on the amount of each ingredient. Recipe retrieval is done considering the ingredients detected, calorie count, various cuisines, and diet types. Users now have an idea of what they can cook, according to the recipes retrieved along with the nutritional value of each recipe. Based on the chosen recipe, similar recipes will be recommended by content-based recommendation system using K-Means Clustering. It helps improve user experience by saving time and energy in finding recipes for daily routines. By retrieving the appropriate recipes based on the items that are accessible and providing precise recipe suggestions, this system simplifies people’s life.