PENGKLASIFIKASIAN TEBU VARIETAS BULULAWANG DENGAN PENDEKATAN DEEP LEARNING (Studi Kasus di PG Kebon Agung, Malang)

Muhammad Iqbal, Muhammad Iqbal (2023) PENGKLASIFIKASIAN TEBU VARIETAS BULULAWANG DENGAN PENDEKATAN DEEP LEARNING (Studi Kasus di PG Kebon Agung, Malang). Other thesis, UPN "Veteran" Yogyajarta.

	Text 1. Skripsi Fulltext_122180077_Muhammad Iqbal.pdf Restricted to Repository staff only Download (15MB)
	Text 2. Cover_122180077_Muhammad Iqbal.pdf Download (66kB)
	Text 3. Abstrak_122180077_Muhammad Iqbal.pdf Download (212kB)
	Text 4. Lembar Pengesahan_122180077_Muhammad Iqbal.pdf Download (453kB)
	Text 5. Daftar isi_122180077_Muhammad Iqbal.pdf Download (319kB)
	Text 6. Daftar Pustaka_122180077_Muhammad Iqbal.pdf Download (166kB)

Abstract

xvii
ABSTRAK
PG Kebon Agung merupakan salah satu perusahaan produksi gula. Bahan
pokok pembuatan gula Sebagian besar berasal dari tanaman tebu. Untuk
mengetahui kandungan dari tanaman tebu dilakukan pengukuran analisis kadar gula
pada nira tebu. Kadar kandungan zat terlarut dalam nira tebu disebut dengan brix.
Penelitian ini bertujuan untuk memprediksi dan mengklasifikasi kualitas brix tebu
dengan menggunakan deep learning. Tebu yang diteliti disini jenis varietas tebu
Bululawang
Penelitian ini menggunakan metode You Only Look Once (YOLO) dan
Convolutional Neural Network (CNN) untuk memprediksi dan mengklasifikasi
data berupa gambar. Selain itu, penelitian ini juga menggunakan metode K-Nearest
Neighbors (KNN) untuk memprediksi dan mengklasifikasikan data numerik pada
masing-masing ruas tebu. Data tersebut memiliki atribut yaitu jarak, panjang, berat,
diameter rerata, keliling, dan kandungan brix.
Berdasarkan hasil prediksi dan klasifikasi menggunakan metode YOLO
didapatkan hasil akurasi TEST tertinggi 0,822 (82,2%), sedangkan menggunakan
metode CNN didapatkan hasil akurasi TEST tertinggi adalah 0,615 (61,5%), dan
berdasarkan metode KNN didapatkan hasil akurasi TEST tertinggi sebesar 0,78
(78%). Hasil prediksi dan klasifikasi menggunakan metode YOLO tersebut
kemudian dilakukan optimasi karena merupakan hasil tertinggi. Optimasi
dilakukan dengan mengubah ukuran pixel, jumlah epoch, dan augmentasi. Setelah
dilakukan optimasi, didapatkan hasil akurasi TEST tertinggi sebesar 0.822 (82,2%).
Hal ini dapat menunjukan bahwa metode YOLO merupakan metode yang paling
cocok digunakan untuk memprediksi kualitas tebu.
Kata kunci: Brix, Qualitas, You Only Look Once (YOLO), Convolutional
Neural Network (CNN), K-Nearest Neighbors (KNN)
xviii
ABSTRACT
PG Kebon Agung is one of the sugar production companies, with the main
raw material for sugar production being predominantly sourced from sugarcane
plants. To determine the content of sugarcane plants, an analysis of sugar content
in sugarcane juice, known as brix, is measured. This research aims to predict and
classify the quality of sugarcane brix using deep learning. The sugarcane studied
here is the Bululawang sugarcane variety
The research utilizes the You Only Look Once (YOLO) and Convolutional
Neural Network (CNN) methods to predict and classify image data. Additionally,
the K-Nearest Neighbors (KNN) method is employed to predict and classify
numerical data in each segment of the sugarcane. The dataset includes attributes
such as distance, length, weight, average diameter, circumference, and brix content.
Based on the prediction and classification results using the YOLO method,
the highest accuracy of 0.822 (82.2%) was obtained for the TEST set. Meanwhile,
using the CNN method, the highest accuracy for the TEST set was 0.615 (61.5%),
and based on the KNN method, the highest accuracy for the TEST set was 0.78
(78%). The YOLO method showed the highest accuracy, and optimization was
performed by adjusting the pixel size, number of epochs, and augmentation. After
optimization, the highest accuracy for the TEST set reached 0.822 (82.2%). These
results indicate that the YOLO method is the most suitable for predicting the quality
of sugarcane.
Keywords: Brix, Quality,You Only Look Once (YOLO), Convolutional Neural
Network (CNN), K-Nearest Neighbors (KNN)

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Brix, Quality,You Only Look Once (YOLO), Convolutional Neural Network (CNN), K-Nearest Neighbors (KNN)
Subjects:	H Social Sciences > HD Industries. Land use. Labor
Divisions:	Faculty of Engineering, Science and Mathematics > School of Engineering Sciences
Depositing User:	Eko Yuli
Date Deposited:	22 Jun 2023 04:05
Last Modified:	22 Jun 2023 04:05
URI:	http://eprints.upnyk.ac.id/id/eprint/36088

Actions (login required)

View Item