Hybrid CNN Model for Detection of Diseases in Leafy Plants

Department of Computer Science and Engineering, Brainware University, Barasat, Kolkata, West Bengal, Pin 700125, India

Abstract

Traditional methods for plant disease detection involve expert inspection, which is both subjective and time-consuming. Convolutional neural networks (CNNs), Extreme Gradient Boosting are combined in this model to increase accuracy in plant disease classification. CNN can extract and learn features from leaf images, while EGB optimise accuracy by extracting patterns. XGBoost stands out with its efficient boosting techniques that combine weak learners into stronger classifiers for enhanced performance. Finally, an ensemble technique combining predictions from both classifiers leverages their respective strengths for optimal plant disease detection, achieving an overall accuracy rate of more than 97.2%.