Towards Sustainable Agriculture: Federated CNN Models for Cucurbit Leaf Disease Detection

This study uses a cutting-edge method that combines federated learning and Convolutional Neural Networks (CNN) to tackle the complex problem of diagnosing Cucurbit leaf damage. The research used information from five customers to create a complete model operating within the scope of five different classes of Cucurbit leaf diseases. The complex Result Analysis, which painstakingly assessed the effectiveness of the federated learning model among customers, was a crucial component of the investigation. For Cts-l (92.23%, 92.49%, 92.51%), Cts-2 (94.49%, 94.56%, 94.55%), Cts-3 (96.93%, 96.94%, 96.94%), Cts-4 (90.35%, 90.39%, 90.40%), and Cts-5 (92.69%, 92.70%, 92.71 %), this analysis produced macro, micro, and weighted averages. The study's equally crucial section revolved around Federated Averaging, a method that cleverly transforms regional data averages into worldwide insights, assuring smooth and reliable knowledge transfer. As a result, this transformation of local to global data via Federated Averaging became a game-changer, improving the model's overall comprehension and ability to diagnose Cucurbit leaf illnesses. The study establishes a promising and novel approach for managing Cucurbit leaf disease by combining various experimental and statistical data, substantiating the effectiveness of federated learning in conjunction with CNN in agricultural applications.