A framework for identification of soybean leaf diseases

Abstract

Agriculture being one of the main occupations in India, its contribution to the Indian economy is expected to be significant. However, statistics show that the contribution made by the agricultural sector to Gross Domestic Product (GDP) is comparatively less. Few main reasons for the decrease in agricultural productivity are adverse climatic conditions and attack due to infections and pests. If these diseases are detected at an early stage, timely measures can be taken to control the losses.The proposed work will help farmers to increase their productivity by implementing automated disease detection. The current work focuses on the detection of soyabean leaf diseases using image processing algorithms. The intention of this proposed work is to contribute help to the farmers regarding the health of the plant using decision support system (DSS). The proposed work uses mobile camera for image acquisition and classifies soyabean leaf images as normal or abnormal. The abnormal leaves are further classified into four disease classes – sun-burn, yellow mosaic virus, grass-hopper attack and leaf blight. Support vector machine (SVM) is used for classification giving an average classification accuracy of about 95.33%. The performance of SVM is also compared with-nearest neighbor (KNN) and artificial neural network (ANN).