NutriMatch: AI – Driven Personalized Meal Recipes based on the Fresh Ingredients’ Detection and User’s Dietary Needs

Abstract

In modern fast-paced lifestyles, maintaining a healthy diet is challenging, contributing to the rise of diet-related diseases and food wastage, particularly fresh produce. Existing digital meal planning systems often lack the robustness to integrate practical ingredient detection with personalized dietary requirements effectively. To address these issues, this paper introduces NutriMatch, a web-based application designed to provide personalized healthy meal suggestions based on user submitted images of fresh fruits and vegetables. NutriMatch integrates an ingredient detection interface where users upload produce images via the browser, and the system returns the predicted ingredient label with a confidence score, for example, ginger with 90%, enabling users to verify detection before receiving tailored recipe recommendations. The system utilizes the MobileNetV2 architecture to classify 36 categories of fresh ingredients, chosen for its efficiency and suitability for web-based deployment. The platform is developed using the Laravel framework with PHP and MySQL for backend management, while the frontend utilizes React for a responsive user interface. Experimental results on the test dataset demonstrate that the model achieves a precision of roughly 92 percent and an F1-score of 0.89, validating the system's ability to facilitate sustainable eating habits and personalized nutrition through artificial intelligence model.