Classification of Mango Leaves Infected by Fungal Disease Anthracnose Using Deep Learning
Pankaj Kumar, Sunidhi Ashtekar, S.S. Jayakrishna, K P Bharath, P. T. Vanathi, Rajesh Kumar M
Abstract
Mango is the fruit of high economic and ecological importance in India, as it exports mangoes in large quantities. More than 1500 mango species are cultivated in India and more than 1000 of them are commercial varieties. Mangoes are highly affected by number of diseases, which hamper its appearance, taste and has huge impact on the economy. Amongst number of diseases, Anthracnose is the most commonly occurring fungal disease that is infecting mango trees in India. It is necessary to have an easy and appropriate method to diagnose this highly infectious fungal disease Anthracnose. It would be easier of mango cultivators to identify this disease beforehand and apply proper medication. This will help in preserving its quality and improving the production. Deep learning technologies, computer vision have dragged lot of research attention over past few years due to its high computation and accuracy to classify variety of fungal and bacterial diseases affecting Mango trees. This paper introduces a novel deep learning convolutional neural network (CNN) architecture to identify Anthracnose disease of mango. A real-time dataset captured in farms of Karnataka, Maharashtra and New Delhi is used for validation. It comprises of the images of mango tree leaves by including both healthy and diseased category. In comparison with other state-of-the-art approaches, the proposed algorithm gives higher classification accuracy of about 96.16%.