Abstract: Agriculture is an important source of livelihood where 65% population is depend on it. The crop loss due to disease is increasing day by day which affects on the quality and productivity of crop. As diseases on the crop are certain, the early disease detection of the crop plays major role to control the loss in agriculture. In the proposed disease detection system, the work is carried out on cotton leaves. Initially the infected region is captured and pre-processed by converting captured RGB image to other color space. During segmentation, leaf as well as diseased part is segmented using Otsu's global thresholding method and different features are extracted such as color and texture with the help of color-co-occurrence method. Finally classification technique is used for detecting the diseases with the help of Multi SVM (Multi Support Vector Machine) classifier.
Keywords: crop, disease, color, feature, Multi SVM.