Disease Classification in Wheat from Images Using CNN

Wheat is a staple crop for large sections of the global population. However, like all crops, a major impediment to its yield and expansion is the prevalence of diseases in wheat that cause significant losses to the crop each year. One of the methods that such losses can be prevented is by identifying and responding to the same appropriately and swiftly. Given that a majority of farmers in developing countries cannot afford specialist help for the same, and the fact that manual identification of disease in wheat is a challenging task to begin with, this problem may be aided by using computer vision methods to identify the diseases, after which the farmer needs to only refer to expert sources for solutions. This can greatly reduce the challenge and costs, as well as increase the availability of specialists to deal with the same. It can also be used as an early detection mechanism for the advent of plant epidemics. This review paper aims to perform a comparative analysis of some of the popular deep learning architectures-namely VGG16, MobileNet, ResNet50-for the problem of disease identification and classification in wheat.