A Comparative Analysis on Various Modified Deep Convolution Neural Networks on Maize Plant Leaf Disease Classification

Early detection of crop diseases is an art to prevent further loss in the yield. It not only prevents further spread but also increases the revenue for the farmers. Leaf is a major part of any plant and its good health is very important for maximum yield. Maize is one of the major food crops in the world and contributes major part in food security. There are various reasons for leaf to get diseases that include pathogens like virus, bacteria, pests and even adverse environmental conditions. The idea of this paper is to take data set of maize leaf images with disease like blight, common rust and gray leaf spot along with healthy leaves, train it on different modified deep convolutional neural networks (DCNNs) and use the trained models to test new leaf images. Deep convolutional neural networks have been modified as per the requirements by introducing new epochs, iterations, activation functions and learning rates. In the proposed work, ten different modified networks have been implemented starting from a simple AlexNet to a more complex network like EfficientNet-b0. In this experiment two different data set have been used, one from a standard data set from Kaggle and another is created manually by capturing images from our neighborhood farm lands at different stage of growth. Finally, a comparative study on different networks with two different data set is carried out.