Prediction of Disease with Severity Measure using Optimized Deep Learning Model for Precision Agriculture

Abstract

The widening disparity between the demand and productivity of maize have concerns for the food industry and farmers. Its vulnerability to diseases such as Turcicum Leaf Blight and Rust significantly diminishes its yield. The manual diagnosis and categorization of these illnesses, together with the computation of disease severity and assessment of crop loss, is a labor-intensive endeavor. Additionally, it requires proficiency in illness identification. Consequently, it is imperative to provide an alternative for automated disease identification and severity assessment. The encouraging outcomes of deep learning models in agriculture inspire researchers to use these methods for disease diagnosis and classification maize cultivation. In this research work a hybrid deep learning system for automated illness diagnosis and its severity prediction is introduced. The proposed model is a hybrid version of Oppositional learning and Crow Search optimization (O-CSO) algorithm to fine tune the parameters of CNN to predict the type of disease and its severity simultaneously with high accuracy. The proposed model is evaluated with real time and annotated datasets to prove its significance.