Emerging Trends in Agritech: Federated Learning CNN-Based Jackfruit Leaf Disease Severity Detection

This paper uses federated learning and a Convolutional Neural Network (CNN) to present a new case study in the categorization of jackfruit leaf disease. Information from four customers helps zero in on the illness severity. This paradigm preserves data privacy and collaborative learning by using federated learning without storing sensitive information in a central location. Part of the task is one that defines mild (2650 %, or roughly one in four), severe (51-75 %, almost two out of three) and critical cases( 76-100 %). In this case, we use a federated learning model whereby data from all four clients are sent up to get an average global model. Locally tail Performance measures such as Accuracy, Precision and Recall, F1-Score, and Support are used in the results. These indicate how well an individual model does at classifying a patient's condition severity. We examined customer performance in terms of Macro, Micro and Weighted Averages to give a balanced view on model capabilities. When compared against the results analysis data, these numbers offered strong validation for our model's prowess. For example, the Macro Average values were between 88.44 % and 92.04 %, meaning that all classes performed equally regardless of how many times they appeared in a passage. The range of these values was 88.42 % to 92.06 %. Results arising from the Micro Average, which are especially effective in situations of class imbalance, perform admirably as well. They average between 88.40 % and 92.04 %. This study constitutes another brick for farm technology. It reveals how federated learning can solve problems of complicated, manyclient data sets for disease classification even while maintaining security and accuracy.