Farming Crop in Health Analytics: Federated Learning CNNs in Banana Leaf Disease Detection

In this study, we propose a cooperative learning approach based on Convolutional Neural Networks (CNN) for banana leaf disease classification. Data sets from four clients will be used to classify the disease into four severity levels. A complete understanding of the disease's progression requires understanding the disease's severity classes. It is crucial for agricultural settings to maintain private and decentralized data processing, which can be achieved through federated learning. There are remarkable results in this study. Using the federated learning model, all clients have a high precision rate, recall percentage, and accuracy for disease classification. The performance metrics for each client are as follows: number 1 achieved a macro average of 92.04%, weighted about Processing local data from individual customers and converting it to global data via federated averaging techniques, resulting in persistent, reliable model performance. This consistent picture is evident in different averaging methods, proving the strength of federated learning. As a result of the study, local data characteristics have been turned into a working global model, confirmed to be accurate and applicable to various agricultural scenarios. Federated learning is an effective tool for agricultural development, allowing farmers to share data locally while maintaining the accuracy of the global model. Based on this data, agricultural decisions can be made regarding crop production, water use, and other aspects of farming.