Backpropagation and self-organizing map neural network methods for identifying types of eggplant fruit

Abstract

Identification of the image of eggplant fruit is a process aimed at obtaining objects contained in the image of eggplant fruit or dividing the image of eggplant fruit into several regions with each object or region that has similar aribut, such as shape, color, and size. The formulation of the problem in this study is how to get the objects contained in eggplant fruit and how to calculate the accuracy of the results of training on Backpropagation artificial neural networks and Self-Organizing Map (SOM) from an image of eggplants in all directions. The purpose of this study is to compare the accuracy of the process of identifying eggplant fruit types with the process of training Backpropagation and Self-Organizing Map (SOM) artificial neural networks based on size using the MATLAB application. After testing with the backpropagation method one eggplant image class can be classified with one eggplant, so the accuracy produced by the backpropagation network in the testing process is 18.18% and with the Self-Organizing Map (SOM) method, so the accuracy is generated by the network SOM in the testing process is 23.18%.