Comparison of ALEXNET and VGG16 for Analysis of Plant Leaf Disease

Abstract

Plant leaves are most commonly affected by diseases caused by bacteria, fungi or viruses which results in an immense decrease in the yield from plants. Since most of the people in India are dependent on agriculture there is a need to detect the plant leaf diseases at an early stage. This paper discusses the plant leaf disease detection using two convolutional neural networks that is AlexNet and VGG16. Both the model was trained using dataset of 38 different classes of plant leaves. The role of number of images, learning rate and freezing of layers in the classification accuracy and training time have been analyzed. Further the prediction of noisy images was performed by using both models and remedy for the disease was displayed.