Claim Missing Document
Check
Articles

Found 2 Documents
Search

Analisis Pengolahan Ekstraksi Fitur Citra Untuk Klasifikasi Jenis Apel Menggunakan Scikit-Learn Dengan Algoritma K-Nearest Neighbor Wariyanto Abdullah, Robi; Kusumastuti , Rajnaparamitha; Handoko
Jurnal Teknologi Informasi dan Ilmu Komputer Vol 12 No 1: Februari 2025
Publisher : Fakultas Ilmu Komputer, Universitas Brawijaya

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.25126/jtiik.2025129149

Abstract

Jenis Apel di Indonesia makin beragam seiring dengan perkembangan teknologi dibidang perkebunan.apel yang berkembang pesat. Penggunaan teknologi pengolahan citra dan pembelajaran mesin telah membuka peluang baru dalam mengatasi tantangan klasifikasi objek berbasis citra, termasuk dalam mengidentifikasi jenis buah apel.Penggunaan metode K-Nearest Neighbor sudah banyak terbukti dalam klasifikasi berbagai jenis data termasuk citra. Penelitian yang akan dilakukan akan mencoba melakukan analisa dan mengklasifikasikan jenis apel dengan membaca extrasi fitur citra menggunakan library scikit-learn dalam bahasa pemrograman Python dengan pendekatan algoritma K-Nearest Neighbor. Dataset yang akan diteli dalam klasifikasi digunakan 7 jenis apel yaitu Apple Braeburn, Apple Crimson Snow, Apple Golden, Apple Granny Smith, Apple Red , Apple Red Delicious, Apple Red Yellow. Dataset training yang digunakan dalam penelitian sebanyak 16404 citra, sedangkan data testing sebanyak 2134 citra apel.  Proses klasifikasi dilakukan dengan membandingkan extrasi fitur dari apel yang belum diketahui varietasnya dengan fitur-fitur dari apel yang telah diklasifikasikan sebelumnya.extrasi fiture yang akan dilakukan yaitu akan membandingkandan dari hasil extrasi fiture HVS, histogram dan RGB yang dipilih dengan nilai k genap. Hasil penelitian yang dilakukan menghasilkan akurasi tertinggi sebesar 96,9 %, dengan nilai k=2 dan kriteria perhitungan jarak menggunakan extraksi fitur menggunakan RGB,HSV dan histogram. Penelitian ini diharapkan mampu memberikan potensi penggunaan teknik pengolahan citra dalam mendukung identifikasi jenis apel secara otomatis, yang relevan dalam industri pertanian dan pengolahan makanan. Penelitian selanjutnya diharapkan dapat dilakukan dengan membandingkan metode algoritma untuk klasifikasi yang lain serta memberikan dataset image dengan pencahayaan yang berbeda dengan menambah beberapa jenis apel  dan kombinasi parameter yang lebih banyak lagi agar dapat meningkatkan output hasil penelitian yang sudah dilakukan.   Abstract Apple types in Indonesia are increasingly diverse along with the rapid development of technology in the field of apple plantations. The use of image processing and machine learning technology has opened up new opportunities in overcoming the challenges of image-based object classification, including in identifying apple types. The use of the K-Nearest Neighbor method has been widely proven in the classification of various types of data including images. The research that will be conducted will try to analyze and classify apple types by reading image feature extraction using the scikit-learn library in the Python programming language with the K-Nearest Neighbor algorithm approach. The dataset that will be studied in the classification uses 7 types of apples, namely Apple Braeburn, Apple Crimson Snow, Apple Golden, Apple Granny Smith, Apple Red, Apple Red Delicious, Apple Red Yellow. The training dataset used in the study was 16404 images, while the testing data was 2134 apple images. The classification process is carried out by comparing feature extraction from apples whose varieties are unknown with features from apples that have been previously classified. The feature extraction that will be carried out is to compare and from the results of the HVS, histogram and RGB feature extraction selected with an even k value. The results of the research conducted produced the highest accuracy of 96.9%, with a value of k = 2 and the distance calculation criteria using feature extraction using RGB, HSV and histograms. This research is expected to provide the potential for using image processing techniques to support automatic identification of apple types, which are relevant in the agricultural and food processing industries. Further research is expected to be carried out by comparing algorithm methods for other classifications and providing image datasets with different lighting by adding several types of apples and more parameter combinations in order to increase the output of the research results that have been carried out.
Automated and Efficient Monitoring System for Organic Waste Compost Processing based on The Internet of Things (IoT) Sugiarto, Lilik; ady saputra, indrawan; Wariyanto Abdullah, Robi
BEST Vol 8 No 1 (2026): BEST
Publisher : Universitas PGRI Adi Buana Surabaya

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.36456/7madhs96

Abstract

In developed countries, waste has been regarded as an important component of management systems as well as reuse practices. In contrast, developing countries, particularly Indonesia, still face various challenges in waste management. Approximately 60% of the total national waste generation originates from household waste, and about 39.98% of this amount has not been optimally managed. Processing organic waste into compost is an environmentally friendly alternative that can reduce waste volume while increasing the value of household and agricultural waste. However, conventional composting methods often encounter difficulties, especially in maintaining temperature and moisture stability, causing the decomposition process to be less optimal. Based on these issues, this study aims to design and implement an automated efficiency and monitoring system for compost processing based on the Internet of Things (IoT). The developed system utilizes an ESP32 microcontroller, a soil moisture sensor for moisture measurement, a DS18B20 sensor for compost temperature monitoring, as well as an automatically controlled water pump and a 12 V DC fan. Sensor data are transmitted in real time to the Blynk platform for remote monitoring purposes. The experimental results indicate that the system is capable of maintaining moisture levels within the ideal range of 50–60% and compost temperature within the optimal range of 30–40°C, enabling the composting process to operate more stably, efficiently, and in a controlled manner.