2) ABSTRAK Gandum adalah jenis tanaman yang kaya karbohidrat. Permintaan gandum di Indonesia selalu meningkat setiap tahun tetapi berbanding terbalik dengan jumlah produksi gandum nasional. Salah satu faktor yang menghambat produksi gandum adalah kegagalan panen akibat penyakit atau hama. Penyakit yang umum pada tanaman gandum adalah Septoria dan Stripe Rust. Penyakit tersebut dapat diidentifikasi melalui warna dan bercak daun. Seiring perkembangan teknologi, petani dapat mengawasi tanaman secara otomatis menggunakan bantuan komputer. Dengan menggunakan deep learning, penyakit pada tanaman gandum dapat diidentifikasi dengan mudah. Penelitian ini bertujuan untuk mengidentifikasi penyakit pada tanaman gandum melalui daun menggunakan metode Residual Network (ResNet). ResNet adalah jenis arsitektur Convolution Neural Network (CNN) dengan menggunakan model yang sudah dilatih sebelumnya. Dengan ResNet tidak memerlukan untuk melatih data dari awal sehingga dapat mempersingkat waktu. Data yang digunakan terdiri dari 291 gambar yang terbagi menjadi normal, penyakit Septoria, dan penyakit Stripe Rust. Setelah pengujian didapatkan akurasi sebesar 98% dengan perbandingan data latih dan uji sebesar 90:10 dan nilai confusion matrix sebesar 0.35 sehingga dapat disimpulkan bahwa metode ResNet dapat mengidentifikasi penyakit pada tanaman gandum.
Diseases that attack banana plants can affect the growth and productivity of the fruit produced. The disease can be identified by looking at changes in the pattern and color of the leaves. Infected leaves will experience an increased transpiration process and the photosynthesis process is almost non-existent. Furthermore, disease on banana leaves can cause yield losses of up to 50%. Therefore, early detection is needed so that diseases on banana leaves can be overcome as soon as possible by using deep learning. This study aims to compare the performance of DenseNet and Inception methods in detecting disease on banana leaves. DenseNet is a transfer learning architecture model with fewer parameters and computations to achieve good performance. Inception, on the other hand, is a transfer learning architectural model that applies cross-channel correlation, executes at lower resolution inputs, and avoids spatial dimensions. In conducting the test, this study uses several data handling schemes to test the two methods, namely without data handling, under-sampling, and oversampling. Furthermore, the data is separated into training data and test data with a ratio of 80:20. The result is that the model using the DenseNet method with an oversampling scheme is superior to other models with a percentage value of 84.73% accuracy, 84.80% precision, 84.73% recall, and 84.62% f1 score. In addition, the machine learning model using the DenseNet method in all schemes is also superior to the machine learning model using the Inception method.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
customersupport@researchsolutions.com
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.