KLASIFIKASI KEMATANGAN BUAH SEMANGKA DENGAN METODE PERCEPTRON
No Thumbnail Available
Date
2022-12
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Elfitra
Abstract
Watermelon is a fruit that is widely consumed by humans, watermelons that are consumed
are ripe, ripe watermelons and raw watermelons tend to have the same skin color, which
is green, so that visually distinguishing the human eye is very difficult to distinguish.
common people. Along with the development of technology, it is possible to regret this
problem, namely the technique of image processing, with the help of digital image
processing. This study aims to build a system that can classify raw and ripe watermelons
using the perceptron method. The perceptron method is a method that is able to perform
calculations by recognizing variables in pattern matching. In this method there are
several variables that must be initialized, namely the input value, weight value, bias value
and learning rate. The application of the perceptron model in this study uses an initial
weight value of 0, an initial bias value of 0 and a learning rate of 1. The data used is a
digital image of the watermelon in JPG format, totaling 100 images, 50 raw and 50 ripe.
50 images are used as training data and 50 images are used as testing data. In a digital
image composed of RGB colors, the warrant value is converted to the HSV color space
and only the hue value is taken. Before taking the hue value, the digital image object is
processed, namely by giving a filter to remove the background so that the color object is
not influenced by other colors. The hue value obtained is then processed using a single
layer perceptron and a bipolar sigmoid activation function. We found a convergent
weight at the 102 epoch with a bias value of -142 and a weight value of 0,586354073.
From this value, testing is carried out and results in 92% accuracy from 50 testing data.
Description
Keywords
Watermelon, RGB, HSV, Perceptron, artificial neural network
Citation
Perpustakaan