An intelligent and fast system for detection of grape diseases in RGB, grayscale, YCbCr, HSV and L*a*b* color spaces

To avoid loss and maintain the quality of the crop, identifying plant diseases is essential. However, it is challenging and time-consuming to identify a plant disease on-site. An intelligent method to identify and categorize diseases in grape plants is described in this work. In the proposed method, digital images of grape plants are used to find the features of healthy and disease-affected plants. Global features determine color and texture information, while patterns or structures (e.g., corners or edges) are detected using the speeded-up robust features (SURF) method. The number of features is reduced by quantifying feature space using the K -means clustering algorithm. The concise set of features serves as the training set of the support vector machine (SVM) classier. During testing, the system determined the class of unlabeled images using the trained SVM classifier. The original data set consists of 1600 images in RGB color space belonging to one healthy and three disease classes Black-measles, Black-rot, Leaf-blight, and Healthy-leaf. Simulations are performed in four color spaces: grayscale, YCbCr, HSV, and L*a*b*, to test the system and compute accuracy and confusion matrices. On-the-fly conversion transforms RGB images to other color spaces. The system achieved a maximum average accuracy of up to 90.63