Multilayer Convolutional Neural Network for Maize Leaf Disease Classification

Abstract

In agriculture, early and accurate detection of diseases of plants leaf is very crucial for healthy growth of crops. After rice and wheat crop, maize is India's third-most significant cereal crop but the quality of maize crop is affected by blight, common rust, gray leaf spot leaf diseases. Since huge amount of manpower and time are needed for manual detection of maize leaf diseases. Authors have proposed automated effective model for classification of maize plant leaf diseases using multilayer convolutional neural networks, which will automatically extract relevant image features and identify plant leaf diseases. Here, online dataset is used which is freely available from Plant village and it consist of plant leaf diseases four classes (blight, common rust, gray leaf spot and healthy). Total 4187 images have been used for classification of disease classes and our model show the classification accuracy of 93.28%. The aim of proposed work is to assist the farmers for accurate disease so that they can take further steps for crop protection using different pesticides suggested by agriculture professionals. This approach will definitely improve the crop yield quality.