Empowering Farmers with AI: Federated Learning of CNNs for Wheat Diseases Multi-Classification
Shiva Mehta, Vinay Kukreja, Satvik Vats
Abstract
Higher agricultural outputs are required due to the rising worldwide population, shifting nutritional preferences, and growing demand for food and basic materials for the industry. However, the farming sector confronts several difficulties, such as climate change and a rise in the severity of production risks, which have had a detrimental effect on food output. Crop production forecast algorithms have been created using machine learning and deep learning techniques to handle this issue. Traditional machine learning techniques, however, are less effective because many characteristics related to meteorological data, earth data, and agricultural management data are dispersed and isolated to specific organization computers or smart farming devices. Using about 9,876 pictures, a collaborative learning CNN method for wheat disease identification is suggested. The suggested method used the federated averaging technique to train the model on laterally dispersed datasets across various client devices. The suggested method beat current state-of-the-art models for detecting wheat illness, according to the findings of our experiments, obtaining high accuracy, precision, recall, and F1 score. The suggested method shows how federated learning can enhance machine learning models in a distributed environment. It can also be applied to other agricultural uses, such as crop forecast, soil analysis, and insect detection.