A Novel-Based Deep Learning Approach for Early Diagnosis of Cotton Crop Disease

Abstract

Pakistan is among those countries where most of the GDP of a country depends on agriculture. As agriculture plays a vital role in the economic stability of Pakistan. The farmers are encountering various challenges in diagnosing diseases in cotton crops for the last decade. The attack of disease ultimately lessens the cotton yield and as a result, the economy of the country and the income of the farmers also go down. The detection of disease in crops at an early stage is crucial which can ultimately save farmers from loss. Manual detection is difficult and time-consuming an automated system to detect disease in plants with high accuracy will be handy for farmers. A deep learning approach is used in the proposed work to detect the disease in the crop plant. The Deep Convolutional Neural Networks model has been utilized to detect cotton plant disease using the image dataset of the cotton plant leaves. The model achieved a training accuracy of 99.08% validation accuracy of 100%. The proposed methodology will allow farmers to correctly detect diseases in the early stage. The diagnosis of plant disease helps to cure it in time. If the disease is controlled in time, its ultimate effect will be on the yield of the cotton crop. This will improve farmer income and the GDP of a country.