Flavour Fusion: AI Recipe Generator

Abstract

Food photography is popular because people enjoy visually appealing meals, yet images alone do not convey how to prepare the dish. This project addresses that gap by using artificial intelligence to generate recipes from food images. It aims to support healthy eating, reduce food waste, and assist in meal planning. We use a hybrid machine learning approach that combines convolutional neural networks (CNNs) for analyzing food images and recurrent neural networks (RNNs) for generating corresponding recipes. Our method improves the accuracy of ingredient recognition and recipe generation by considering local cuisine styles and food presentation. We created a custom dataset of 100 South Indian dishes for training and evaluation. The model produces step-by-step instructions based on the visual features of the dish. The proposed system achieves an accuracy of 98%, making it a reliable and efficient solution for recipe generation.