A New Diagnostic Approach for the Detection of Wheat Leaf Disease Using Deep Transfer and Ensemble Learning Based Models

More than one third of the global population consume wheat as a staple food. However, the production of this vital crop is hampered by various challenges, with diseases being a chief concern. Various factors, spanning from climate and soil to pests and pathogens, influence wheat health. Among the numerous diseases affecting wheat, leaf rust stands out as a severe disease hampering both the quality and quantity of wheat produced. In order to combat these diseases, research scholars are utilizing cutting-edge tools like deep learning and image processing to identify and categorize wheat plant diseases, thereby increasing crop yields and reducing losses. Recent studies have closely studied the practical viability of these advanced techniques in the context of wheat plant disease detection and classification. This research introduces a state-of-art framework for wheat disease detection. By harnessing the power of transfer-learning and ensemble-learning, the methodology achieves an outstanding testing accuracy of 99.56%. It performs remarkably better than popular deep learning models like VGG19, ResNet50, and DenseNet121. Importantly, the experimental results highlight not only the method's higher performance in key measures including precision, recall but also its elevated accuracy. This research's breakthrough holds promise in addressing the pressing issue of disease detection in wheat cultivation, offering a brighter outlook for effective crop management and increased production.