Classification of Orange Fruit Using Convolutional Neural Network, Support Vector Machine, K-Nearest Neighbor and Naive Bayes Methods Based on Color Analysis

Abstract

Citrus fruit is a fruit that has good vitamins and is popular with the public. This fruit also has various types with different benefits. Each type of orange also has a variety of colors. Types of oranges can be checked manually by looking directly at the color and texture of the fruit. This manual method is very simple but also very subjective because of the different understanding of each person about the types of oranges. Therefore, this research discusses and explains how to determine the type of fruit by comparing several methods, namely using the SVM method (Support Vector Machine), the CNN method (Convolutional Neural Network), the K-NN method (K-Nearest Neighbor), and the Naïve Bayes method by taking several samples of citrus fruit images consisting of sweet oranges, lemons and limes using a mobile phone camera. The total dataset used in this study is 90 datasets consisting of 30 sweet orange images, 30 lime images and 30 lemon images. Of the 90 datasets are divided into training data and test data. From the results of the study, it was obtained that the accuracy of compatibility with a percentage of 100% using the CNN method (Convolutional Neural Network).