A Comparative Study of Classifiers in the Context of Papaya Disease Recognition

Nowadays, machine learning techniques have been effectively being applied to a wide area of applications. Although a large number of state-of-the-art classification algorithms have been applied in different applications, they are infrequently tested in the same classification problem domain. In this paper, nine (9) prominent classification algorithms are compared in index of six (6) performance metrics in a computer vision context. Machine vision based papaya disease recognition can help to build an online agro-medical expert system that recognizes the defects of fruit by diseases from an image that is taken using mobile or another handheld device in order to distantly help both beginner and professional growers in the agriculture-based country like Bangladesh. In this context, since a classifier is required, the merits of prominent classification algorithms need to be thoroughly assessed. So, we compare the performances of SVM, C4.5, naive Bayes, logistic regression, kNN, random forest, backpropagation neural network, counterpropagation neural network, and RIPPER classifiers. SVM outperforms all other classifiers achieving more than 95% accuracy, whereas kNN performs worst showing 71.11% accuracy.