Techniques for Plant Disease Diagnosis Evaluated on a Windows Phone Platform

Abstract

The recognition of plant diseases is a responsibility of professional agriculture engineers. Intelligent systems can assist plant disease diagnosis in the early stages with low cost. User descriptions and image comparison are exploited in some expert systems that are already available. More sophisticated techniques like the one presented in this paper are based on features extracted from the symptoms (e.g., lesions) of a plant disease that appear on the leaves, the fruits, etc. The color, the dimensions and the number of these lesion spots can be used in some cases to discriminate the disease that has mortified a plant. In this paper, we describe a smart phone application that measures the features of the plant lesions with higher than 90% precision. The accuracy in the recognition of grapevine or citrus diseases that have been used as case studies is higher than 70% in most of the cases using only 5 photographs for the definition of each disease. The most important advantage of the proposed method is that the set of the supported diseases can be easily extended by the end-user.