Please use this identifier to cite or link to this item:
https://repository.uksw.edu//handle/123456789/36160
Title: | Penggunaan YOLOv8 untuk Deteksi Penyakit Daun Kopi |
Authors: | Bitra, Marcelino |
Keywords: | YOLOv8;Deteksi Objek;Penyakit Daun Kopi;Kecerdasan Buatan |
Issue Date: | Oct-2024 |
Abstract: | Salah satu hasil perkebunan dengan peranan cukup penting dalam kegiatan perekonomian di Indonesia adalah kopi. Tapi, produksi kopi di Indonesia mengalami penurunan, dimana salah satu penyebabnya adalah serangan hama dan penyakit. Kecerdasan buatan dapat menjadi solusi untuk membantu petani mendeteksi penyakit pada tanaman kopi dengan menggunakan algoritma deteksi objek. Penelitian ini menggunakan algoritma deteksi objek YOLOv8 untuk melakukan deteksi pada keadaan dan penyakit daun tanaman kopi yang dibagi ke dalam empat klasifikasi, yaitu miner, rust, phoma dan healthy. Penelitian dilakukan dalam tiga skenario percobaan yang dibedakan dengan berdasarkan pada perbandingan pembagian data pada train set, validation set, dan test set, dimana secara berurutan dari train, validation, dan test, skenario pertama memiliki perbandingan 80:10:10, skenario kedua 70:15:15, dan skenario ketiga 70:20:10. Proses penelitian menggunakan model YOLOv8s mendapatkan model dengan hasil performa terbaik pada perbandingan data 70% train set, 20% validation set, dan 10% test set. Model dengan performa terbaik memiliki nilai mAP sebesar 97,8%, precision 95,2%, recall 96,6%, dan f1-score 96%. One of the products of plantation with a significant role in economic activities in Indonesia is coffee. But, coffee production in Indonesia is experienced a decline, where one of the causes is pest and disease attacks. Artificial intelligence can be a solution to help farmers detect diseases in coffee plants using object detection algorithm. This research uses the YOLOv8 object detection algorithm to carry out detection of the state and diseases of coffee plant leaves which are divided into four classifications, namely miner, rust, phoma and healthy. The research was conducted in three experimental scenarios which were differentiated based on a comparison of data distribution in the train set, validation set, and test set, where in sequence of train, validation, and test, the first scenario had a comparison of 80:10:10, the second scenario 70: 15:15, and third scenario 70:20:10. The research process using the YOLOv8s model got a model with the best performance results in data comparison of 70% train set, 20% validation set, and 10% test set. The best performing model has a mAP value of 97.8%, precision 95.2%, recall 96.6%, and f1-score 96%. |
URI: | https://repository.uksw.edu//handle/123456789/36160 |
Appears in Collections: | T1 - Informatics Engineering |
Files in This Item:
File | Description | Size | Format | |
---|---|---|---|---|
T1_672020304_Judul.pdf | 851.38 kB | Adobe PDF | View/Open | |
T1_672020304_Isi.pdf Until 9999-01-01 | 597.24 kB | Adobe PDF | View/Open | |
T1_672020304_Daftar Pustaka.pdf | 253.92 kB | Adobe PDF | View/Open | |
T1_672020304_Formulir Pernyataan Persetujuan Penyerahan Lisensi dan Pilihan Embargo.pdf Until 9999-01-01 | 784.45 kB | Adobe PDF | View/Open |
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.