Image Features Based Intelligent Apple Disease Prediction System: Machine Learning Based Apple Disease Prediction System

Image Features Based Intelligent Apple Disease Prediction System: Machine Learning Based Apple Disease Prediction System

Mahvish Jan (Central University of Jammu, Jammu, India) and Hazik Ahmad (Sher-e-Kashmir University of Agricultural Sciences and Technology of Kashmir, India)
DOI: 10.4018/IJAEIS.2020070103


A pattern classifier (PC) is used to solve a variety of non-separable and complex computing problems. One of the key problems is to efficiently predict a type of disease in a typical fruit tree. The timely and accurately predicted disease in an apple tree may help a farmer to take appropriate preventive measures in advance. In this article, an apple disease diagnosis system is developed to predict the apple scab and leaf/spot blight diseases. In this article, low level and shape-based features are used for the development of an intelligent apple disease prediction system. First, the key image features like entropy, energy, inverse difference moment (IDM), mean, standard deviation (SD), perimeter, etc., are extracted from the apple leaf images. The model for the proposed system is trained by using multi-layer perceptron (MLP) pattern classifier and eleven apple leaves image features. The Gradient descent back-propagation algorithm is used for building the intelligent system to carry out the pattern classification. The proposed system is tested using some random samples and exhibits excellent diagnosis accuracy of 99.1%. The sensitivity of the proposed prediction model is 98.1% and specificity of ~99.9%.
Article Preview


Kashmir is famous for its specialty of apples all over India. India contributes 65% of total production of apples in the world. Like other fruits, apple (Malus pumila) is also attacked by a number of diseases such as apple scab and Alternaria leaf spot/blight. These diseases can affect both the leaves and fruit which may amount loss in both the quality as well as quantity (Kayalvizhi and Antony, 2011). Apple Scab disease is caused by the fungus Venturia inaequalis. It infects leaves, shoots, buds, blossoms and fruit. It is the most serious and prevalent disease. At the initial stage, the symptoms usually appear as small spots on the lower side of leaves or as spots on either surface of older leaves. The disease causes pale yellow or brown or olive-green color on the upper surface of the leaves. The lesions on older leaves are more definite in outline and become velvety grey to sooty black. The lesion may form a convex surface with corresponding convex area on the lower side. If infection is severe, leaves will become twisted, distorted and stunted and tend to fall prematurely. Alternaria leaf spot/blight is caused by a fungal pathogen Alternaria Mali. The lesions of this disease in apple first appears on leaves in late spring and early summer as round, small, purplish or blackish spots. Initially the leaf spots are 1/8 to 1/4 in diameter. The disease spreads rapidly over a temperature between 77 to 86°F (25 to 30°C) and due to wet conditions. At ideal temperatures infection occurs with 5.5 hours of wetting, and lesions can appear in the orchard two days after infection causing a serious epidemic. The main objective of this research work is to predict the apple disease by using the low-level image features of the leaves of an apple tree. The texture or statistical features of leaves of a healthy tree and other one with diseases are distinctive which may be used to train a machine leaning model. The automatic prediction of diseases in apple trees in the Kashmir valley of India has been addressed by this research work. The contribution of the work is that it achieves better prediction accuracy for apple disease by using only low-level texture or shape features of apple leaf images. The system mainly focuses with a new dataset of apple leaf images collected afresh from Kashmir region of India where 75% of the total apples are grown in India.

The rest of the paper is organized as follows. The section two of the paper provides a brief background about machine learning based solutions for disease prediction in various fruits and vegetables. The third section of the article provides the detailed architecture and algorithm used for our proposed apple disease prediction system. The next section describes about the experimental setup and results analysis for the proposed system. The last section provides a comparison of the proposed technique with the existing similar techniques with brief conclusion.

Complete Article List

Search this Journal:
Open Access Articles
Volume 12: 4 Issues (2021): Forthcoming, Available for Pre-Order
Volume 11: 4 Issues (2020): 3 Released, 1 Forthcoming
Volume 10: 4 Issues (2019)
Volume 9: 4 Issues (2018)
Volume 8: 4 Issues (2017)
Volume 7: 4 Issues (2016)
Volume 6: 4 Issues (2015)
Volume 5: 4 Issues (2014)
Volume 4: 4 Issues (2013)
Volume 3: 2 Issues (2012)
Volume 2: 2 Issues (2011)
Volume 1: 2 Issues (2010)
View Complete Journal Contents Listing