An IoT based System with Edge Intelligence for Rice Leaf Disease Detection using Machine Learning
S. M. Shahidur Harun Rumy, Md. Ishan Arefin Hossain, Forji Jahan, Tanjina Tanvin
Abstract
Bangladesh is one of the top five rice-producing and consuming countries in the world. Its economy dramatically depends on rice-producing. Rice leaf disease is the biggest problem in the agriculture sector. This is the main reason for the reduction of the quality and quantity of the crops. The spread of the disease can be avoided by continuous monitoring. However, manual monitoring of diseases will cost a large amount of time and labor. So, it is a good idea to have an automated system. This paper presents a rice leaf disease detection system using a lightweight Artificial Intelligent technique. We are applying the edge computing concept here. Our edge device is Raspberry Pi. We have processed all our data in Raspberry Pi. We consider three rice plant diseases, namely Brown Spot, Hispa, and Leaf Blast. They are the most common type of rice leaf disease in Bangladesh. We have used clear images of healthy and infected rice leaves with white background. After applying the necessary preprocessing, we have extracted the necessary features from the images. Then we have made an image classification model with various machine learning algorithms by feeding these features. We have learned that the Random Forest algorithm performed the best. By using our image classification model, we have achieved 97.50% accuracy on our edge device.