Improved Plant Disease Detection Techniques using Convolutional Neural Networks: A Survey

Abstract

Predicting and detection of plant disease has always been a very serious problem faced in agricultural field. People with low financial backgrounds cannot bear the expenses of regular checkups of their crops and fields. Thus, a technique which can fulfill the needs of the poor farmers will be game changing invention. The proposed model in this paper has come forward with a system capable of detecting whether a plant is healthy or not. Further if the plan is unhealthy then the cause of the disease is also identified via taking two inputs such as plant leaves and soil sample from the plot where the diseased plant is present. The model is trained in order to classify whether the disease found is bacterial or fungal or due to pest attack. If the plant doesn’t show any symptoms of being infected then the report of soil nutrient content is too generated and collectively provides the final result. The result will further be accompanied by recommending the required fertilizer or pesticide to tackle the problem and thereby reducing the loss in the production.