CROP DISEASE DETECTION USING DEEP LEARNING

Abstract

—This research presents an innovative application of deep learning techniques for the earlydetection and classification of plant diseases, highlighting the exceptional performance of theResNet9 model. The study begins with a meticulous implementation of ResNet9 on the diverse andextensive PlantVillage dataset, achieving an impressive 99.2% accuracy. This success is attributedto strategic parameter tuning, incorporating techniques such as learning rate scheduling, gradientclipping, and weight decay. The investigation is expanded to address unique challenges in Indianagriculture, with the curation of datasets for major crops such as cotton, rice, and groundnut. Thecommitment to practical application is manifested in the development of a userfriendly webinterface, strategically designed to empower farmers. This interface, based on Convolutional NeuralNetworks (CNNs), facilitates accurate disease identification across different crops, offering acomprehensive solution for precision agriculture. Through this tool, farmers gain valuable insightsinto disease prevention methods, enhancing their decisionmaking in sustainable crop healthmanagement. The findings underscore the efficacy and versatility of the approach, positioning it atthe forefront of leveraging technology for the advancement of global food security and agriculturalsustainability.