Zea Mays Leaf Disease Classification using Swin Transformer

Agricultural losses have an effect on the global economy, and plant diseases are a primary cause of such losses. To address these challenges, artificial intelligence techniques such as deep learning can help to leverage the consequences. Corn is one of the most significant agricultural products. A disease outbreak might cause a significant drop in corn production, resulting in millions of dollars in losses. The risk of crop failure due to a disease pandemic can be mitigated with the help of deep learning methods. Traditionally, plant illnesses are examined using just one’s own eyes, with the emphasis typically being on color changes, the presence of spots or rotten regions in the leaves, or both. Because of their tiny nature, the symptoms of plant diseases might be difficult for farmers to accurately detect in certain cases. In addition, many people who operate farms are not specialists in the identification and classification of diseases; thus, vision-based deep learning methods may be able to aid farmers in delivering more accurate diagnoses, identification, and classification. Convolutional neural networks have emerged as the top choice for image processing over the last few years as advancements have been made in related domains. Swin Transformer, which was recently introduced, has shown significant improvement in classification applications. We used the Swin Transformer to categorize maize leaf diseases including blight, common rust, and grey leaf spot. Compared to the state-of-the-art approaches currently available, our model has the highest accuracy at 95.9%.