Apple Ripeness Level Detection Based On Skin Color Features With Convolutional Neural Network Classification Method

Abstract

Apple ripeness detection system based on skin color image is a system that can aid apple farmers, sellers, and buyers to determine the maturity or ripeness level of an apple based on its skin color. One of the weaknesses of identifying the level of ripeness of apples is the use of conventional methods using human power to assess the ripeness level of apples. This method certainly very tentative towards different perspectives of every human being. In this study, a ripeness detection system build for Manalagi Apples and Rome Beauty Apples was designed based on skin color images using the Convolutional Neural Network classification method trained with Deep Learning to predict the maturity level classification. From the training and testing during classification, showed that Manalagi Apples and Rome Beauty Apples that are predicted in the form of three ripeness level, such as: Ripe, Half Ripe and Raw, had very good accuracy with an average prediction accuracy rate of 86-95%.