Please use this identifier to cite or link to this item: http://repository.ipb.ac.id/handle/123456789/112152
Title: Model Deep Learning YOLO untuk Deteksi Keberhasilan Tanam Padi (Oryza sativa) berbasis Citra Unmanned Aerial Vehicle
Other Titles: Deep Learning Model YOLO for Detecting the Success of Rice (Oryza sativa) Planting Based on Unmanned Aerial Vehicle Image
Authors: Supriyanto
Purwansya, Yuvicko Gerhaen
Issue Date: 2022
Publisher: IPB University
Abstract: Deteksi keberhasilan tanam padi (Oryza sativa) pada fase awal budidaya perlu dilakukan secara ketat agar dapat dimanfaatkan untuk kegiatan penyulaman, pengendalian hama, pengendalian penyakit, prediksi kebutuhan pupuk, dan perlakuan pemeliharaan tanaman yang tepat. Citra tangkapan dari Unmanned Aerial Vehicle (UAV) sangat potensial digunakan untuk mengamati lahan padi yang luas dengan bantuan algoritma kecerdasan buatan. Penelitian ini bertujuan mengembangkan model deep learning untuk mendeteksi tegakan padi di sawah pada fase awal budidaya menggunakan citra RGB beresolusi tinggi hasil tangkapan UAV. Penelitian ini menggunakan deep learning dengan algoritma YOLOv5 yang terdiri dari tahapan: (1) pengumpulan dataset, (2) pelabelan dataset, (3) training dataset, (4) uji model hasil training, dan (5) implementasi model. Training deep learning terhadap 9.677 dataset dilakukan dengan menggunakan algoritma YOLOv5s, YOLOv5m, dan YOLOv5l. Akurasi dari model deep learning menggunakan algoritma YOLOv5s, YOLOv5m, dan YOLOv5l masing-masing sebesar 89,35%; 91,29%; dan 79,43%. Nilai presisi yang didapatkan sebesar 99,90%; 99,80%; dan 99,89%. Sedangkan nilai recall yang didapatkan sebesar 89,43%; 91,46%; dan 79,50%. Hasil tersebut menunjukkan model deep learning yang telah dikembangkan mampu mendeteksi tegakan padi dengan akurasi tinggi dan presisi. Model dapat digunakan untuk mendeteksi padi pada fase awal budidaya dengan user interface berbasis web yang diimplementasikan dengan bahasa pemrograman Pyhton pada komputer lokal.
Detection the success of rice (Oryza sativa) planting in the early stages of cultivation are important for embroidery activities, pest control, disease control, prediction of fertilizer needs, and proper plant maintenance treatment. The captured image from the Unmanned Aerial Vehicle (UAV) coupled with artificial intelligence are potential to observe large scale of rice fields. The objective of this study is to develop a deep learning model to detect rice stands using high resolution RGB imagery captured by UAV in the early stages of paddy. Deep learning with the YOLOv5 algorithm was used to develop the model that consists of the following stages: (1) dataset collection, (2) dataset labeling, (3) dataset training, (4) model testing, and (5) model implementation. Deep learning was employed to train 9,677 datasets using YOLOv5s, YOLOv5m, and YOLOv5l algorithms. The accuracy of the deep learning model using the YOLOv5s, YOLOv5m, and YOLOv5l algorithms was 89.35%; 91.29%; and 79.43% respectively. The precision value obtained is 99.90%; 99.80%; and 99.89% respectively. While the recall value obtained was 89.43%; 91.46%; and 79.50% respectively. Based on these results indicate that the deep learning model was capable to detect rice stands with high accuracy and precision. The model can be used to detect rice in the early stages of cultivation with a web-based user interface implemented with the Python programming language on a local computer.
URI: http://repository.ipb.ac.id/handle/123456789/112152
Appears in Collections:UT - Agricultural and Biosystem Engineering

Files in This Item:
File Description SizeFormat 
Cover, Lembar Pernyataan, Abstrak, Lembar Pengesahan, Prakata dan Daftar Isi.pdf
  Restricted Access
Cover679.58 kBAdobe PDFView/Open
F14180014_Yuvicko Gerhaen Purwansya.pdf
  Restricted Access
Fullteks1.33 MBAdobe PDFView/Open
Lampiran.pdf
  Restricted Access
Lampiran1.15 MBAdobe PDFView/Open


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.