<![CDATA[Classifying the quality of corn seeds by manual visual observation takes a long time. It also produces products with uneven quality due to visual limitations, fatigue, and differences in observer perception. This research aims to classify superior corn seeds using the machine learning method, namely K-Nearest Neighbors (K-NN). The research data uses 500 images of corn seeds consisting of 400 training images and 100 test images. Extraction of corn image features uses the Gray Level Co-occurrence Matrix (GLCM) method to obtain texture characteristics. The texture characteristic values of metric natural corn images concist of contrast, energy, homogeneity and correlation. Based on the image texture characteristic values, classification is carried out using the K-Nearest Neighbor (K-NN) method. The classification results consist of classes of viable and non-viable corn seeds. The performance evaluation metric method calculates accuracy, sensitivity and specificity using a confusion matrix. This research shows that the value of k=5 is the most optimal, and the accuracy, sensitivity and sensitivity values, respectively, are 75%, 77% and 72% found in the ninth fold]]>
Copyrights © 2024
