AI-Powered Wheat Disease Detection: A CNN-Logistic Regression Framework

Early diagnosis and timely intervention in the wheat leaf’s case, infection guarantee crop health and yield. This study describes classification based on deep learning model to detect precisely seven wheat leaf infection: Fusarium Leaf Blight, Yellow Leaf Spot, Blanqueta, Leaf Spot, Wheat Rust, and Yellow Leaf Rust. The model’s overall accuracy is $97.46 \%$, and the database contains 12,405 photos. The performance of this model is gauged by Preciseness, Accuracy, Support, F1-Score, Support Proportion, and Recall. Outcomes indicate that high classification accuracy was obtained, with Stem Rust having a Preciseness, F1-score, and Recall of 97.63%, Stripe Rust, Septoria Leaf Blotch, and Fusarium Leaf Blight being 97.41%, 97.45%, and 97.55%, respectively. Likewise, Tan Spot, Powdery Mildew, and Spot Blotch received F1 scores of $97.26 \%, 97.60 \%$, and $97.28 \%$, demonstrating the strength of the model. Accuracy-wise, every class of diseases attained a 99% accuracy, testifying to the credibility of the model. Support for the classes of diseases varied between 1645 for Tan Spot and 1895 for Stem Rust with a support ratio ranging from 0.13 to 0.15 to avoid an imbalance of the data. The suggested model offers a very accurate and effective solution for wheat leaf disease detection, allowing for timely intervention and enhancing disease management strategies for sustainable agriculture.