Detecting Jackfruit Leaf Disease with Advanced Federated CNN Models Brings New Frontiers

The use of convolutional neural networks (CNNs) in the context of federated learning is presented in this paper in order to introduce a new method for disease classification on the jackfruit leaf. Using data from five customers, we train the model to target five different types of jackfruit leaf diseases. Using decentralized data, federated learning keeps information private and reduces data transfer. Based on our review, the model appears to be quite successful in different ways. A macro average, a micro average, and a weighted average are used to measure efficacy based on local (client-specific) and global (aggregated) data. Customer IR-1 averages 93.35 %, customer IR-2 averages 92.08 %, and customer IR-2 averages 75.64 %. Ir-i through ii had weighted averages of 93.34 %, 93.09 %, 94.18 % and 57 respectively (with adjustments reflecting differences in relative amounts between various expectation levels); this statistic indicates how well a model can handle common diseases by identifying the class frequency within the data. The Weighted Average from overall accuracy was almost identical to that of Micro-Average; customers' scores for were 93.34 %, 97 %,81 %. Based on these four numbers, the model shows high accuracy rates across all classes, demonstrating its overall effectiveness. We improve the robustness and generalizability of the system for classifying illnesses through Federated Averaging, simplifying transmission from local to global data studies.