<ul><li>Crop disease detection and classification is a crucial challenge in agriculture, impacting productivity, food security, and environmental sustainability.</li><li>Deep learning models like CNN and ViT excel in classifying plant diseases but are restricted due to data privacy concerns.</li><li>A new Decentralized Federated Learning (DFL) framework is introduced, using validation loss to guide model sharing and correct local training.</li><li>Experiments with PlantVillage datasets and various deep learning architectures show that DFL enhances accuracy, convergence speed, generalization, and robustness across diverse data environments for privacy-preserving agricultural applications.</li></ul>

Loss-Guided Model Sharing and Local Learning Correction in Decentralized Federated Learning for Crop Disease Classification

Discover more