Transforming Okra Cultivation: A Study on Leaf Disease Using Federated Learning CNNs

Convolutional neural networks (CNNs) with federated learning are used in this study paper's innovative classification method for five different Okra leaf diseases. The research is being conducted in a federated learning architecture and comprises five clients, each specializing in a specific type of illness. It aims to solve issues with data protection and accessibility. In the study of the results, the model performs well in terms of precision, recall, F1-score, and accuracy metrics. By illustration, Client 1 achieves a mean accuracy and recall of 93.76% and 93.75%, respectively, while Client 2 stands out with a remarkable mean precision of 96.35% and a mean recall of 96.10%. With accuracy, memory, and F1 scores over 89% and 91%, respectively, even the customers with comparatively lower metrics, customers 4 and 5, provide strong outcomes. The research also proposes federated averaging, which keeps performance high while transforming local data into global data. The macro, micro, and weighted averages all demonstrate the model's outstanding performance in this area; Client 2, for example, achieved macro, micro, and weighted averages of 96.22%, 96.47%, and 96.47%, respectively. Overall, the study highlights how well CNNs trained in a federated learning framework perform at diagnosing Okra leaf illnesses, displaying excellent efficiency even when converting local data to global data. This novel strategy shows potential for real-world, valuable applications in the agricultural sector, especially in areas with a strong Okra farming industry.