utomatic Recognition of Avocado Fruit Diseases using Modified Deep Convolutional Neural Network

Abstract

Fruit diseases are extremely significant which affects the quality and quantity of fruit
production. Avocado fruit is one of the most required food sources in the world owing to its rich
nutrition. The diseases of avocado fruit degrade the yield of the crop and cause economic loss to the
farmers. The fruit is vulnerable to diseases such as phytophthora, scab, stem end rot, anthracnose, rat
bite, seed moth, sun bloch, and cercospora spot. Therefore, early detection of diseases is very much
essential to take preventive measures to control the spread of diseases. The disease identification
employing traditional methods is time consuming and often requires agricultural expertise. However,
to address this problem, a transfer learning based convolution neural network model is developed to
perform automatic detection of diseases. The MobileNet model is modified by applying the transfer
learning concept which provides better recognition accuracy of 96% on real time images. The
significant success rate makes this model as the early warning tool for disease identification