Classification and Analysis of Tomato Species with Convolutional Neural Networks

Abstract

Tomatoes are one of the most used vegetables. There are varieties that can grow in different climates. The taste, usage area and commercial value of each are different from each other. For this reason, identifying and sorting tomato species after the production stage is a problem. In addition, since tomato is a sensitive vegetable, it is extremely important to separate it from a distance. For this purpose, the classification of tomato images belonging to 9 different tomato species was carried out in the study. In total, a dataset containing 6810 tomato images in 9 classes was used. Three different pre-trained Convolutional Neural Network (CNN) models were used with the transfer learning method to classify the images. AlexNet, InceptionV3 and VGG16 models were used for classification. As a result of the classifications made, the highest classification belongs to the AlexNet model with 100%. Evaluation of the performances of the models was also made with precision, recall, F1 Score and specificity performance metrics. It is foreseen that the proposed methods can be used for the separation of tomatoes.