A Novel Maize Plant Disease Classification using Federated Machine Learning in Smart Agriculture

This study aims to investigate the ability of Federated Machine Learning (FML) with ResNet50 to detect illnesses in maize leaves. We conduct a comparative analysis to evaluate the efficacy of three convolutional neural network (CNN) models in a decentralized environment, contrasting with conventional centralized approaches. The objective is to assess the efficacy of FML in safeguarding data privacy while simultaneously achieving excellent classification performance, as evaluated by accuracy, precision, recall, and F1-score. Additionally, the study explores the advantages of utilizing ResNet50 in terms of accuracy and decreased training time. The study emphasizes the importance of implementing suitable preprocessing, augmentation methods, and classification techniques using FML with CNN models ResNet50, SqueezeNet, and AlexNet, particularly for maize crops. We deploy FML + ResNet-50, FML + AlexNet, and FML + SqueezeNet models to assess their practical applications. By incorporating FML + ResNet-50 fusion into plant classification software systems, farmers can optimize decision-making and boost maize disease classification, acquiring geographical context, improving classification accuracy, and ensuring data privacy. In machine learning, deep learning, and CNN, FML + ResNet50 significantly surpassed the performance of current centralized categorization methods. ResNet-50 with FML did better than other classification methods on a public dataset, with 96.49% accuracy, 96.46% precision, 96.39% recall, and 96.40% F1-score. Research Originality: This study has not been previously published.