A Robust And Novel Multilayer Convolutional Neural Network For Classification And Validation Of Plant Leaf Disease

Abstract

The core of the Indian economy is agriculture, yet the persistent problem of plant disease detection jeopardizes its productivity. Plant diseases can significantly lower the quality and quantity of agricultural goods, which has a severe effect on food production safety. Plant diseases can potentially completely prevent grain harvests in extreme circumstances. In the realm of agricultural informatics, the automatic identification and diagnosis of plant diseases is therefore widely desired. Numerous approaches have been put out to tackle this task, with deep learning emerging as the go-to approach because of its remarkable results.

As a result, a Multilayer Convolutional Neural Network (MCNN) is suggested for this work in order to classify the disease-affected plant leaves. This work has been verified using a real-time dataset that was Offline augmentation . This dataset, which is divided into 38 classes, includes over 87K rgb photos of both healthy and damaged crop leaves. The entire dataset is split up into training and validation sets in an 80/20 ratio while maintaining the directory structure. For prediction purposes, a new directory with 38 test photos is later established. Findings indicate that, in comparison to other cutting-edge methods, the suggested MCNN model has a better classification accuracy.