Disruptive Tech in Agriculture: Federated Learning CNNs for Soybean Leaf Disease Classification

This research paper shows a groundbreaking study about sorting soybean leaf disease severity by feeding learning with convolutional neural networks (CNN). The study looks closely at soybean leaf disease and groups it into four levels (1-25%, 26–50%, 51–75%, and 76–100%) in four clients. It incorporates local details to create a model that can be adapted worldwide. The method employed in federated learning collects information from different local sets of data (showing various environmental conditions and sickness expressions) while still keeping it private. Using CNN in this group setup allows us to classify disease severity quickly and correctly. This method works well with different data types and gives a complete view of how disease affects people. Looking at the four clients, we found significant differences in how well the models did. This is shown by the numbers that measure accuracy, like the success rate and finding all items correctly or wrongly ignored. It also shows how successful they are. The significant, small, and heavy averages were calculated for each customer. This provided a detailed look at how appropriate the approach was at different levels of seriousness. Client MP_5 showed better results with 96% accuracy. This was followed by Client mp_4 at 87%, Client mp_6 at 88%, Client mp_3 at 82%, and Client mp_1 to be named. However, these extra costs can degrade the quality or gameplay experience of some popular multiplayer games. In short, the study shows that federated learning and CNN can work well for classifying soybean leaf diseases. The mix of local knowledge and worldwide use sets an unprecedented standard for studying farm data and planning disease management in agriculture.