A Deep Learning Approach for the Classification of Rice Leaf Diseases

Abstract

The fast and appropriate analysis and recognition of plant diseases can control the growth of diseases on various crops towards improving the quality and productivity of crops. The automatic system can perform disease recognition at minimum cost and error without the farm specialist’s interpretation. It is very difficult to manually identify appropriate properties for distinguishing different kinds of crop diseases by using image processing and machine learning methods. In this study, we have developed a convolutional neural network (CNN) framework, a deep learning approach for automatically classifying three kinds of rice leaf diseases such as bacterial blight, blast, and brown mark. In the first phase, the developed system distinguished healthy and diseased leaves from a set of 1500 rice leaves. In the second phase, the three kinds of diseases have been categorized from a dataset containing 500 images of each of the three kinds of diseased rice leaves. The CNN model automatically learned required properties from raw images to differentiate the healthy and diseased rice leaves with 94% accuracy and then categorized different kinds of diseased rice leaves with 78.44% accuracy.