A Machine Learning Approach for Vegetable Recognition

Abstract

In perceiving objects in an image, computer vision and example recognition is an emerging territory. The advances in perceiving objects in pictures have a wide range of applications, including frameworks for discovering vegetables and natural products, vehicles, and other frameworks. Our research focuses on the discovery of vegetable varieties in order to create a reliable vegetable recognition system. Since the vegetables can appear the same due to shading and different highlights, such as red tomato and red capsicum having similar tones, using highlights to differentiate vegetables can result in false discovery, so this research proposes an intraclass vegetable recognition system based on profound learning. Three forms of vegetables were used. Those are Carrots, Tomatoes and Cauliflowers. By extracting and learning the images, profound learning had been used to classify the vegetable class, and the convolution neural network (CNN) was investigated. Intraclass vegetables were viewed accurately with 95.50% accuracy and proficiently using profound learning, according to the results of the evaluation.