Ensemble Learning Model for Fruit and Vegetable Classification

Abstract

Nutritional information about fruits and vegetables is vital for promoting healthy eating patterns and combating malnutrition. This research presents the development of a web application for fruit and vegetable image classification, utilizing ensemble learning with a stacking technique. The model combines Swin Transformer and ResNet as base learners, with Support Vector Machine (SVM) serving as the meta-learner. Trained on a dataset encompassing 32 fruit and vegetable classes, the model achieved an impressive 98% accuracy, along with consistently high precision, recall, and F1-score. The application was implemented with Flask for the backend and ReactJS for the frontend and is hosted on PythonAnywhere. Beyond image classification, the application provides users with detailed nutritional information, including energy content and vitamin composition, in a quick and user-friendly manner. This study highlights the effectiveness of ensemble learning in enhancing classification accuracy. Future work will focus on expanding the dataset and transitioning to more robust hosting platforms to improve performance and user experience.