Early Detection of Cercospora Cotton Plant Disease by Using Machine Learning Technique

Abstract

Agriculture is known as the backbone of every nation. Cotton is considered an essential plant according to the textile point of view for making cotton cloths, bed sheets, towels, handkerchiefs, and many other products. This research's objective is early cotton disease detection by using image processing techniques and finding the disease through an automatic way instead of seeing them manually. Cercospora Leaf Spot (CLS) is a severe problem that may decline cotton production. About 80-95% of the diseases in cotton leaves are similar to Alternaria, Erythema, Leukoplakia, and Macula on leaves. The K-means clustering algorithm is used for the segmentation of image into clusters. The extraction of features is done by using the hybrid method for texture and color feature extraction. Finally, Support Vector Machine (SVM) is used for the classification of Cercospora cotton leaves. At the end, precision and recall performance evaluation metrics are used to evaluate the accuracy and it is found that about 96% accuracy is achieved.