Transformative Diagnostics: The Rise of Federated Learning CNNs in Papaya Leaf Disease Detection

There has been a growing need for practical answers to the problem of papaya leaf diseases. Because of this, to present a new, spread-out, all-encompassing way to diagnose that uses the growing power of federated learning along with convolutional neural networks (CNN). Using a five-client system (cy_1–cy_5), the data was divided into five unique categories of Papaya leaf diseases and analysed using each client's local data. Several statistical breakthroughs were brought to light via extensive training and validation. Precision values of 80.32% for cy_1-fe_1 and an incredible 90.37% for cy_5-fe_5 were revealed in the Result Analysis of local data, with further values for recall, F1-score, support, and accuracy validating the resiliency and flexibility of the federated CNN model. The research took a more in-depth look at the big picture using Federated Averaging methodologies, emphasising Macro, Micro, and Weighted averages to transform local observations into a universal understanding. While other clients maintained vital metrics like 80.46 per cent (cy_1 Macro) and 78.93 per cent (cy_4 Micro), cy_5 demonstrated unrivalled superiority with figures like 83.10% in Macro, 83.76% in weighted and 83.64% in micro averages. Regarding the crunch of translating regional information into global trends, federated Averaging was indispensable. Harmonising client-specific insights into a generally applicable model, such as cy_1's 80.70% accuracy and cy_5's impressive 83.39% recall. In short, the study provides a unique diagnosis tool for papaya leaf diseases. It establishes a precedent for harnessing decentralised data to address domain-specific difficulties using federated learning and CNNs.