Comparison of Pine Seed Quality Classification Using the Naive Bayes and KNN Methods

Abstract

Pine seeds are the seeds of pine trees, which are a type of open-seeded plant known as gymnosperms. Gymnosperm plants have seeds that are not protected by fruit, unlike flowering plants (angiosperms). Pine seeds are typically found inside hard cones. Pine seeds possess several distinctive characteristics, including their small, flat shape and are often equipped with thin wings that aid in their dispersal when released. The process of selecting pine seeds for planting must adhere to established standards of seed quality to enhance desired attributes such as color, texture, and shape in seedlings. Suitable pine seeds for use in planting or propagation are those in a new condition. Quality pine seeds cannot be distinguished by visual inspection alone; alternative tools are required. Given the challenge of differentiating seeds suitable for primary propagation, the researcher proposes a comparison of Pine seed classification using two different methods: the Naïve Bayes Method and K-Nearest Neighbors (KNN). This is expected to enable the accurate detection of pine seeds. The feature extraction method used is the Gray-Level Co-occurrence Matrix (GLCM). The dataset used consists of 165 pine seed samples, comprising 55 images of fresh pine seeds, 55 images of dry pine seeds, and 55 images of decayed pine seeds. Between the two methods, K-NN exhibits the highest percentage value compared to the Naïve Bayes method in the k-fold cross-validation, achieving an accuracy of 95%.