<![CDATA[Manual determination of sweet corn maturity levels still relies on visual observation, which may lead to subjectivity and inconsistency. This research aims to develop a classification model for sweet corn maturity levels based on digital images using the Support Vector Machine (SVM) algorithm with Grid Search hyperparameter optimization. The dataset consists of 600 sweet corn images divided into three classes: immature, optimally mature, and fully mature. The research stages include preprocessing, feature extraction, dataset splitting, SVM implementation, hyperparameter tuning, and model evaluation. Feature extraction results show that all images were successfully processed with a 100% success rate. Testing the SVM model under several dataset splitting scenarios revealed that the 90:10 ratio achieved the highest accuracy of 88.33%. Hyperparameter optimization using Grid Search produced the best parameter combination, namely C = 100, gamma = 1, and Radial Basis Function (RBF) kernel, with a validation accuracy of 89.44%. The optimized model achieved an accuracy of 90.00%, with the best performance in the fully mature class, reaching 95% precision, 100% recall, and a 98% F1-score. The results indicate that the combination of HSV and GLCM features along with hyperparameter optimization can enhance the performance of SVM in classifying sweet corn maturity levels.]]>
Copyrights © 2026
