Deep Learning Approach for Tomato Leaf Disease Detection

Abstract

In Vietnam, where agriculture is the main source of income for the majority of the population, effectively combating crop diseases and increasing crop yield are very important. Plant diseases can cause significant damage to agricultural productivity and product quality. Early detection of diseases could minimize losses for agricultural sector, thereby fostering tangible benefits for rural communities and overall economic. This paper introduces an approach, leveraging from the VGG-19 architecture, to detect plant leaf diseases by analyzing images of crop leaves. The approach was tested on a dataset comprising approximately 18,000 tomato leaf samples. The model was designed to automatically learn important features from tomato leaf images and classify them into different disease categories. Experimental results show that the model achieved a classification accuracy of 93% on the test set. In addition, after building the prediction model, we has also developed an application that allows users to quickly identify diseases on tomato leaves by capturing or uploading images to the application.