Using Deep Learning for Plant Disease Detection and Classification

Abstract

In India's economy, farming is crucial, making early detection of plant diseases an important task. This helps in reducing crop damage and preventing the diseases from spreading further. Numerous plants, such as corn, tomatoes, and potatoes, display evident symptoms of diseases on their leaves. These noticeable patterns can be employed to accurately predict the diseases and facilitate prompt intervention to reduce their impact. The customary method involves farmers or plant pathologists visually inspecting plant leaves and identifying the specific disease. This project involves a deep learning model designed for classifying plant diseases, utilizing CNNs for their proficiency in image classification. The model, which utilizes architectures like MobileNet, InceptionNet, ResNet, and ResNeXt, delivers faster and more accurate predictions than traditional manual methods. Notably, ResNeXt, with its added dimension of cardinality that aids in learning more complex features, achieved the highest accuracy, reaching 98.2%.