Enhanced Corn Seed Variety Detection Using Hybrid Features and Support Vector Machines

Abstract

The accurate classification of corn seed varieties is crucial for agricultural productivity, seed quality control, and genetic research. Traditional methods of seed classification often rely on manual inspection, which is time-consuming, labor-intensive, and prone to human error. In this study, we propose an intelligent classification system based on Multinomial Naive Bayes (MNB) and Support Vector Machines (SVM) for distinguishing corn seed varieties using hybrid features extracted from multiple data sources. The hybrid features combine morphological, color, and texture characteristics obtained from high-resolution images, as well as spectral features derived from near-infrared (NIR) spectroscopy. The proposed approach involves preprocessing the image and spectral data, extracting relevant features, and then applying both the MNB and SVM classifiers for variety prediction. The Multinomial Naive Bayes algorithm is chosen for its simplicity, efficiency, and effectiveness in handling high-dimensional data with discrete features, while SVM is employed for its robustness in managing complex, non-linear relationships in the data and its ability to achieve high classification accuracy with optimal hyperplane separation. A comparative analysis is conducted to evaluate the performance of both classifiers. This study highlights the effectiveness of combining multiple feature types and leveraging both MNB and SVM classifiers for intelligent corn seed variety classification. The proposed method offers a scalable, automated, and reliable solution for seed classification, which can aid in enhancing crop yield, ensuring seed purity, and supporting precision agriculture practices.