Leafy vegetables identification through image processing with transfer learning methods

Abstract

Maintaining a healthy and robust lifestyle necessitates the consumption of nutritious food, with vegetables playing a pivotal role. Among the different types of vegetables, leafy vegetables are one of them which provide a rich source of important nutrients, including vitamins, minerals, and antioxidants. However, it might be a difficult task to correctly identify and distinguish between various leafy vegetables, especially when they have a similar appearance. The challenges can be addressed by replacing the monitoring system with an automated system for leafy vegetable Identification. The aim of this study is to propose a leafy vegetable identifying system using a transfer learning model. A sophisticated machine learning method called transfer learning is used to improve the performance of recognition and classification utilizing image processing. The proposed model focuses on using a pre-trained CNN model to extract leaf attributes and Analysis different kinds of leafy vegetables and identify the actual name of leafy vegetables. In this study, a custom dataset is used for implementing the transfer learning model utilizing image processing which consists of six types of leafy vegetables, including Colocasia, Red Spinach, Malabar Spinach, Jute Spinach, Organic Spinach, and Stem Amarnath. Evaluating the models on the custom dataset, InceptionV3 achieved the highest accuracy which is 99.75% based on performance metrics such as precision, recall, and f1-score. Thus, this study reduces the eye dependency of humans and consumes the time of the general public.