Early and late blight disease identification in tomato plants using a neural network-based model to augmenting agricultural productivity.

Abstract

Computer-advanced technologies have a significant impact across various fields. It is widely recognized that diseases have a detrimental effect on crop productivity and can significantly impact the economy, particularly in agricultural countries. Tomatoes hold great economic importance among cash crops, second only to potatoes. Globally, tomato production reaches a staggering 160 million tons annually, making it even more crucial for agricultural development. Unfortunately, the tomato crop is susceptible to several diseases, with early blight and late blight as two prominent culprits responsible for a production decrease of around 79%. Traditional disease detection and identification methods are time-consuming, expensive, and destructive, often requiring pathologists' expertise. Thus, the primary research objective is to enhance disease identification accuracy by leveraging deep learning techniques. A model based on the inception-V3 architecture has been devised to classify diseases affecting tomato plant leaves. The model was trained and tested using the PlantVillage dataset, which comprises 6000 sample images of tomato leaves. The training and testing process utilized an 80 : 20 ratio, resulting in an impressive classification accuracy of 97.44% for the proposed model. The proposed solution aims to enable the tomato industry to thrive in the global market by mitigating the impact of tomato leaf diseases. By reducing the prevalence of these diseases, the solution can increase demand and contribute to the industry's growth.