Classification of pomelo leaf diseases using convolution neural network

Abstract

Pomelo is an important export fruit of Thailand. However, it is a plant that is susceptible to many diseases. The objective of this work is to identify diseases using the convolution neural network, which most of the diseases of Pomelo indicate on its leaves. Three types of pomelo leaves images including healthy leaves, greening disease, and citrus leafminer are addressed in this work. Transfer Learning techniques based on GoogLeNet, AlexNet, and Squeeznet are utilized for building the proper machine learning model for classifying Pomelo diseases. Image process techniques are also applied to enhance the performance of the model such as edging, grayscale, and rotation. Experimental results show that all three models give a similar performance where GoogLeNet has a bit better performance than that of AlexNet and Squeeznet.