International Journal of Research and Innovation in Applied Science (IJRIAS) | Volume V, Issue VI, June 2020 | ISSN 2454-6194

Implementation of Improved Machine Learning Techniques for Plant Disease Detection and Classification 

 Ibrahim M. Adekunle
Department of ICT, Osun State University, Nigeria

Abstract−Agricultural production plays a big role in economic growth especially in developing countries; one of the biggest problems for a farmer is different type of diseases that affect crop on farm. This problem has destroyed plenty of crops which eventually led to shortage in annual agricultural production across the world. Quality and high production of crops could be determined by early detection of diseases in the crop. Diseases detection through manual observation can be somehow tedious, complex, expensive, and difficult, and subject to rigorous analysis. Although some researchers have worked in this field but most of the existing methods for solving problems in diseases detection have not been effective in terms of real time basis. This paper therefore presents an effective method for efficient detection of maize leaf diseases. The proposed method uses image processing techniques for the extraction of important features in order to showcase the characteristics properties of the image that could be used for the identification. Machine learning techniques are applied to classify the extracted features to separate diseased plant leaf from healthy ones. The experimental results show that the application of modified machine learning techniques could be effectively used for the classification of plant leaf diseases even with an accuracy of 96.7%. This approach would be very useful to farmers to prevent damages of crops, shortage of food production in the society and wasting of money on agricultural products like pesticides and so on.
Keywords−Maize leaf disease, feature extraction, binary images, histogram equalization and machine learning

Read More