A study on plant disease classification using an ensemble of deep learning models

Abstract

This project presents a comprehensive study on the classification of plant diseases using an ensemble of deep learning models. Timely and accurate identification of plant diseases is critical for maintaining crop health and maximizing agricultural productivity. Leveraging pre-trained convolutional neural networks (CNNs) such as VGG16, ResNet50, and DenseNet121, this research aims to construct a robust ensemble classifier. The study investigates the influence of different activation functions (sigmoid and softmax) in the final layers of individual models and explores various ensemble combination rules (majority voting, average, maximum, and product). The proposed approach is evaluated on two diverse plant disease datasets: New Plant Disease Dataset and Tomato Leaf Disease Dataset. Experimental results demonstrate the effectiveness of the ensemble model in achieving superior classification performance compared to individual models, underscoring the potential of deep learning ensembles in advancing plant disease diagnostics.