Discrimination of radish seed cultivars using fluorescence spectroscopy and advanced machine learning algorithms

The differentiation of seed varieties plays a vital role in modern agriculture, since seed type and quality directly influence germination rate, plant vigor, crop yield, and ultimately, farmers’ profitability. In recent years, the integration of decision support systems with advanced analytical tools has become increasingly important to ensure accurate and practical solutions for farmers. Within this scope, the present study addresses the classification of radish seed varieties, namely French breakfast, Nacional 2, Espresso F1, and Red large, by combining fluorescence spectroscopic techniques with state-of-the-art machine learning algorithms. The rationale behind the proposed approach lies in the complementary strengths of both methods: fluorescence spectroscopy provides a non-destructive, rapid, and sensitive characterization of seeds, while machine learning algorithms enhance the ability to recognize subtle spectral differences and achieve reliable classification outcomes. In this study, four classifiers: Fine Tree, Quadratic Support Vector Machine (SVM), Fine k-Nearest Neighbor (KNN), and Neural Networks were employed to evaluate their performance on fluorescence spectral data. The findings revealed that all applied algorithms produced satisfactory classification accuracies exceeding 90%, thereby confirming the robustness of the proposed framework. Notably, the Quadratic SVM model achieved the highest performance with an accuracy of 100%, demonstrating its superior capability in distinguishing radish seed varieties with complete precision. These results highlight that the synergy between spectroscopic data and machine learning models can be effectively utilized in practice for agricultural decision support. The developed methodology offers farmers a practical, non-invasive, and highly accurate tool for radish seed variety identification, which may significantly contribute to improved seed management, resource optimization, and sustainable agricultural practices.