Support Vector Machine-based Early Detection of Pearl Millet Diseases for Improved Crop Yield

This research outlines a foundational framework for the identification of pearl millet leaf diseases, employing a combination of handcrafted image features and a Support Vector Machine (SVM) classifier. The methodology involved processing a dataset structured into healthy, rust, and downy mildew categories. To ensure consistency and computational efficiency, all images were standardized in size, followed by preprocessing steps including contrast enhancement and color space conversion to optimize the visual prominence of disease symptoms. Subsequently, handcrafted features, designed to capture specific visual patterns associated with each disease type, were extracted. An SVM model was then trained on these features to accurately distinguish among the defined categories. Performance metrics revealed an overall accuracy of 88%, complemented by strong class-wise F1-scores: 93% for healthy, 84% for rust, and 86% for downy mildew, indicating dependable performance across all classes despite any inherent data imbalance. Further analysis through precision and recall metrics confirmed the model's efficacy in correctly identifying both diseased and healthy samples. This work provides a robust baseline, highlighting the capabilities of traditional machine learning in plant disease detection and paving the way for future integrations of more sophisticated techniques for sustainable agricultural applications.