Detection of Rice Plants Diseases Using Convolutional Neural Network (CNN)

Abstract

As a rice-producing plant, rice plant (Oryza sativa L.) is one of the most important crops in Indonesia. Rice production is increasing every year along with an increase in rice demand and population.The amount of rice production is affected by the condition of the rice plants. The worse the condition of rice plants, the rice production will also lower. Rice plant is very susceptible to diseases or pests that can reduce its productivity, including brown spot disease, leaf smut and bacterial leaf blight. As the development of science and technology, currently known as Artificial Intelligence. Artificial intelligence is a combination of several scientific disciplines such as mathematics, statistics, computer science, and even social science. Using artificial intelligence, the system now have the ability to interpret external data correctly to learn from the data and then use the learning to achieve certain goals through flexible adaptation. The artificial intelligence fields consists of several branches, such as machine learning and deep learning. Neural Network (NN) is one of the methods used in the deep learning.NN has many types, one of which is the Convolutional Neural Network (CNN). CNN is the best-knownmethod used for processingimages data compared to other types of NN. Therefore, in this study the identification of rice plants diseases was carriedout using CNN method. From this study,better results were obtained compared to other methods, obtaining 100% accuracy for training data and 86,67% for testing data. The model obtained by the CNN method can be used for detecting 3 different types of rice plants diseases, there are brown spots, leaf smuts, or bacterial leaf blight disease based on the physical images of rice plant leaves.

Penerapan Metode Convolutional Neural Network (CNN) Dalam Mengklasifikasikan Penyakit Daun Tanaman Padi

“…Penelitian keempat dilakukan oleh (Rasjava et al, 2020) yang berjudul "Detection of Rice Plants Diseases Using Convolutional Neural Network (CNN)". Penelitian ini memberikan hasil yang lebih baik dibandingkan dengan metode lain, memperoleh akurasi 100% untuk data pelatihan dan 86,67% untuk data pengujian.…”