Agricultural Disease Detection: A Federated CNN Framework for Strawberry Leaves

This study looks at a novel way of applying federated learning and convolutional neural networks (CNN) to find diseases in strawberry leaves. The study puts the problem at four levels of seriousness and uses four different people. The main goal is to use a learning method called federated learning. This way, you can learn from different ways of displaying data without risking your privacy. Strawberry leaf disease has four levels of seriousness: mild (1-25%), moderate (26–50%), severe (51–75%), and critical (76– 100%). The study shows that the model works well for four customers. It has positive measures like accuracy, recall, and F1-score while focusing on support and precision. Importantly, the big and small averages for every customer show that the model works effectively at different levels of seriousness. For example, Client jm_1 had a big average score of 93.79 and small average scores as well, which are also around the same levels, like 93.74 and 93.75, where weight matters in these cases but it's not much different from each other here so they didn't really notice any significant kind of gap between all their measurements they took at those Also, Client jm_2 had an overall average of 91.95 when looking at all results together. It also had a weighted average of 91.84 and a detailed or close-up version that was 91.85 on more specific details in the study reports they have prepared over time for their use in dealing with customers online via digital marketing. Clients jm_3 and jm_4 had substantial average scores of 93.73 for each, small top-weighted averages at 91.46 together, as well as tiny under-weighted averages around the same amount—that is, their own individual rates.