A Machine Learning Approach for Detection of Cotton Leaf Disease

Cotton is the most important crop in the world and provides the raw material to the textile industry. The main problem faced by farmers in cultivation of the cotton crop is the attack of bacterial diseases, fungal viruses, and attack of worms or unmonitored cultivation lead to crumpling of leaves. So the farmers need to know the diseases attacked on the crop and take necessary measures to avoid poor yield. The traditional methods of prediction of the diseases are not accurate as the experts may get wrong with prediction. The experts use their previous knowledge and provide solutions which are less accurate. So the proposed methods such as image processing and machine learning techniques are used to detect and classify cotton leaf diseases. Comparison of the classification performance is done for several inputs, the system gives an accuracy of 70% for Multi class support vector machine, 86% for K-Nearest Neighbor classifier and for Artificial Neural Network the accuracy achieved is 89%.The project is currently designed to detect three types of diseases which can be further used to find other diseases that may affect various crops.