Potato Leaf Disease Detection using Image Processing with Transfer Learning

Abstract

According to the Indonesia Central Bureau of Statistics, the province of North Sumatra produced about 1309976 tons of potato in 2019-2020. Limitation of farmers to detect plant diseases may cause crop failure. Plant diseases commonly found in potato leaves, such as late blight and early blight, might affect the quality of the potatoes. Early detection of leaf diseases can help farmers prevent further plant damage. This research presents a Convolutional Neural Network model with ResNet-50 architecture that can do image processing with a minimum accuracy of 85%. The Dataset used in this research was from Plant Village, which contains 2152 images and is classified to each disease. This research achieved the model accuracy of 98.4%, according to trials done.