Smart Deep System to Detect and Diagnose Wheat Disease

Abstract

Wheat is the most important basic component of food security in addition to its impact on the agricultural and economic income of countries, although every year we notice significant losses in it, and one of the most important reasons for these losses is the diseases and fungi that infect it, including those who are difficult to distinguish or classify and how to deal with them. The technological development that the world is witnessing in the field of industrial intelligence has become touching various fields, including agriculture, where we began to see today what is called ―Smart agriculture‖. Our project involves the establishment of a smart system that verifies and classifies wheat diseases (brown rust, yellow rust, leaf rust, sptoria) using images processed by convolutional neural networks (CNNs). The first stage is dedicated to a neural network we called the ―Master‖ or DDN1 (Detection Disease Network), which determines whether the wheat is diseased or not. At the end of this stage, we will have one of two results: either the wheat is healthy, concluding the first stage, or the wheat is diseased and that need to call the second system that we called the ―Slave‖ or DDN2 (Diagnose Disease Network), prompting the second stage, where the specific wheat disease is classified. Therefore, this program can be considered a foundation for the construction of the smart system, in addition to providing a strong basis for the future development of agricultural control systems, which will enhance production and surplus food security.