Модель оцінювання якості фруктів

Abstract

This study focuses on the identification of fruit ripeness through visual cues. The primary objective is to achieve high prediction accuracy and ensure portability for deployment on smartphones. To accomplish this, the system employs two machine learning models: one for classification and another for grading. Additionally, visual guidelines are incorporated to assist users in scenarios where the AI models may encounter difficulties. Throughout the development process, various state-of-the-art approaches including deep learning and visual transformers were explored. The system comprises several components, including a mobile application, an API server, a database, and servers housing the classification and grading models.