An Efficient Diagnostic Approach for Multi-Class Classification of Wheat Leaf Disease Using Deep Transfer and Ensemble Learning
Sudhir Saraswat, Salil Batra, Protyush P Neog, Emani Likith Sharma, P. Pavan Kumar, Ankit Kumar Pandey
Abstract
Wheat (Triticum aestivum L.) is a critical component of global food security, feeding billions of people worldwide. However, the high occurrence of diseases in wheat crops poses a significant threat to agricultural production. For efficient management and the protection of food supply, prompt and accurate detection of these diseases is essential. Deep learning has been a powerful method for automating the diagnosis of illnesses affecting wheat leaves in recent years. There are over dozens of wheat diseases that can harm the crop. As a result, manual diagnosis of these diseases can be very challenging. Automated disease classification in wheat can help improve both crop yield quantity and quality. It can also be a helpful tool for crop quality assessment and pricing. The proposed research provides a comprehensive analysis of recent advances in deep learning-based methods for wheat leaf disease diagnosis and a state-of-the-art methodology has been proposed to accurately detect 8 different classes of wheat leaf disease. The proposed framework performs significantly better than popular deep learning models such as CNN, SimpleNet, EfficientNet, VGG16, ResNet50, and VGG-FCN-VD16 by harnessing the power of transfer learning and ensemble learning. Experimental results show that the proposed method also outperforms all existing models with a classification accuracy of 98.08%.