An Enhanced Wheat Stripe Rust Segmentation Approach Using Vision Transformer Model
Abstract Worldwide, the wheat industry encounters major obstacles caused by stripe rust disease, triggered by the fungus Puccinia striiformis. This disease results in considerable losses of wheat crops and has significant economic repercussions. Accurate detection of crop diseases is vital for susta...
Saved in:
| Main Authors: | , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Springer
2025-06-01
|
| Series: | International Journal of Computational Intelligence Systems |
| Subjects: | |
| Online Access: | https://doi.org/10.1007/s44196-025-00873-w |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Abstract Worldwide, the wheat industry encounters major obstacles caused by stripe rust disease, triggered by the fungus Puccinia striiformis. This disease results in considerable losses of wheat crops and has significant economic repercussions. Accurate detection of crop diseases is vital for sustainable agriculture and food security. By effectively identifying and managing crop diseases, yield losses can be prevented and global food production can be ensured. This not only protects farmers’ livelihoods but also contributes to human health and well-being by safeguarding the food supply. Wheat plays a crucial role in Pakistan’s agriculture, covering 37% of the cultivated land and contributing 70% to total production. Stripe rust, a serious fungal disease, heavily affects wheat yield, causing a global loss of 5.5 million tons annually. Current models exhibit shortcomings in accurately detecting wheat stripe rust, necessitating improvements for more precise identification and diagnosis of the disease. Fortunately, recent advances in deep learning have led to significant improvements in object-detection accuracy, thus offering hope for better disease management. The purpose of this study is to introduce a method to minimize losses by accurately and promptly detecting stripe rust disease, thereby avoiding the need for manual inspection. To achieve this goal, we propose a vision transformer and a hybrid model to identify wheat stripe rust by analyzing multi-spectral and high-resolution image data. Additionally, we utilize two models, ViT-Base/16 and a transformer, which prove to be highly effective in accurately detecting diseases using the same datasets as the vision transformer model. The proposed method provides optimal results with a vision transformer and a hybrid approach with 98% accuracy and 97.9%, respectively. ViT-Base/16 and transformer models achieve an accuracy of 98% and 95%, almost. Using an improved vision transformer, we achieved precise detection of wheat stripe rust compared to previous methods. This has various benefits, such as safeguarding yields, reducing costs, supporting sustainable agriculture, facilitating crop monitoring, and improving disease management. |
|---|---|
| ISSN: | 1875-6883 |