Maize Foliar Fungal Disease Detection and Classification Using Convolution Neural Network

Abstract

Three prevalent maize leaf diseases common rust, gray leaf spot, and northern leaf blight lead to significant financial losses for the global maize business. An early and precise diagnosis of a disease can prevent financial losses, lower the use for pesticides, and increase maize output and food security. When there is a large amount of training data available, deep learning techniques, such as convolutional neural networks (CNNs), offer accurate, efficient, and automatic diagnosis on server platforms. The convolutional neural network (CNN) architecture used in this study is trained on a local dataset of images of maize leaves infected with different fungal diseases. The trained CNN model is then used to classify new maize leaf images into healthy or diseased categories and further classify them into specific fungal diseases. The primary objective of this thesis is to develop an accurate and efficient system for the detection and classification of foliar fungal diseases of maize using deep learning techniques. In the proposed architecture, we split the work of maize foliar fungal disease detection and classification into two steps. The first step can be detecting maize foliar fungal disease by training a detection model to detect the disease. A classification model in a Convolutional Neural Network (CNN) architecture is the second step to classify the disease type detected by the detection model. To improve the execution of the network with small computational complexity EfficientNet-B7 is utilized as a feature extractor in the proposed architecture. Moreover, the Accuracy, Precision, and Recall scores of the proposed solution along with different experiments have been measured and the proposed model improves the Accuracy value by 99.25 %, Precision value 99.0 %, and Recall value 98.75 %. Finally, from the evaluation results, the added EfficientNet-B7 has improved the quality of the image being reconstructed.