A Deep Learning Framework for Precise Detection and Classification of Wheat Leaf Diseases

Millions of people depend on wheat as a staple food, especially in agrarian nations like Bangladesh. It is a crop of global importance. Many foliar diseases, such as Septoria Tritici Blotch (STB), a fungal infection that causes tan lesions and yellow halos, pose a threat to its productivity. Manual inspection for traditional disease diagnosis is labor-intensive, prone to mistakes, and not scalable. Recent developments in deep learning and image processing provide a promising substitute for highly accurate automated plant disease detection. With an emphasis on Septoria, this study suggests a thorough deep-learning framework for the identification and categorization of wheat leaf diseases. The methodology entails gathering high-resolution images of wheat leaves from public and research institutions. The images are subjected to color-based and threshold segmentation to isolate infected regions following initial preprocessing, which includes noise reduction, enhancement, and standardization. After that, thirteen texture features that represent color and structural patterns are extracted using the Gray-Level Co-occurrence Matrix (GLCM) technique. Multiple classification models, such as Random Forest (RF), Support Vector Machine (SVM), k-nearest Neighbors (k-NN), and Naïve Bayes (NB), are then trained and assessed using these features. Python and the TensorFlow, Keras, and Mahotas libraries are used to implement the system. Confusion matrices are used to calculate performance metrics like accuracy, sensitivity, specificity, precision, and error rate. Based on experimental results, the Random Forest classifier performed better than the others, achieving 98.9% accuracy, 100% precision and specificity, and 98.1% sensitivity. This validates the suggested method's resilience in comparison to conventional classifiers. The results point to the possibility of implementing deep learning-based technology in actual agricultural environments, supporting sustainable farming, yield enhancement, and early disease detection.