A Novel Transformer Using Dynamic Range-Enhanced Discrete Cosine Transform (DRE-DCT) for Detecting Bean Leaf Diseases

Ibrahim Furkan Ince¹Harisu Abdullahi Shehu^2*Shaira Osmani³Faruk Bulut^4,5*

• ¹Istinye University, Istanbul, Türkiye
• ²Victoria University of Wellington, Wellington, New Zealand
• ³Women in Tech, Washington, United States
• ⁴Istanbul Esenyurt University, Istanbul, Türkiye
• ⁵University of Essex, Colchester, United Kingdom

Early detection of diseases on bean leaves is critical for preventing reduced agricultural productivity and mitigating broader agricultural challenges. Some bean leaf diseases are difficult to detect even with the human eye, presenting significant challenges for machine learning methods that rely on precise feature extraction. In this study, we propose a novel method for detecting bean leaf diseases by combining a preprocessing technique, the dynamic range enhanced discrete cosine transform (DRE-DCT), with Transformer models, referred to as DCT-Transformers. The DRE-DCT method enhances the dynamic range of input images by extracting high-frequency components and subtle details that are typically imperceptible, while preserving image quality. Transformer models were applied to classify bean leaf images both before and after preprocessing. Experimental results show that the DCT-Transformers method achieved a classification accuracy of 99.56% (precision-0.9916, recall-0.9912, and F1-score-0.9912) with preprocessed images, compared to 95.92% with non-pre-processed images. Furthermore, the DCT-Transformers method outperformed state-of-the-art methods (all below 94%) and similar studies (all below 98.5%). These findings highlight the importance of enhancing feature extraction to improve classification performance through preprocessing, offering an efficient solution for disease management in agriculture and ensuring food security.