Solanaceae Family Plants Disease Classification Using Machine Learning Techniques
Gaurav Pandey, Rajneesh Sharma, Amit Kukker
Abstract
The Solanaceae family's tomato and potato crops are particularly vulnerable to early blight and late blight. These two diseases greatly reduce the crop yield. Here, authors consider potato crop samples for rapid and accurate diagnosis of potato leaf diseases. Visual inspection is the most challenging task for farmers for deworming. In this paper, for the diagnosis of potato plant leaf diseases, two automatic detection methodologies are proposed. In the first approach, gray level co-occurrence matrix (GLCM) and 2-D discrete wavelet transform applied to extract features from leaf images and support vector machine, logistic regression, and K-nearest neighbors machine learning techniques are applied for classification. In second approach, convolutional neural networks are applied to identify diseases. The findings show that the convolutional neural networks-based model beats its competitor techniques in various evaluation parameters. The convolutional neural networks based model achieves the best accuracy 97%, followed by SVM 93 %, logistic regression 91 % and KNN 89%. The proposed method will undoubtedly assist farmers in identifying crop disease and take timely decisions for protection.