<![CDATA[Indonesia is the world's largest palm oil producer, with annual production reaching more than 45 million tons. The quality of oil palm fresh fruit bunches (FFB) determines the quality of the oil produced. The quality of FFBs can be seen through their maturity and health. Fruit that is not ripe, overripe, or contaminated with mold can reduce oil quality due to high levels of free fatty acids (FFA). This research aims to build a classification model of FFB marketability using the K-Nearest Neighbors (K-NN) algorithm with RGB and GLCM features. Image data was collected from the plantation, then processed through the stages of preprocessing, feature extraction, and normalization. The model was tested in three approaches, namely using RGB-GLCM combination features, RGB only, and GLCM only, with various data sharing scenarios, namely 70:30, 80:20, and 90:10, as well as varying k values, namely k = 3, 5, 7, 9. The evaluation results show that the RGB-GLCM feature combination model in the 80:20 data sharing scenario and k = 5 value is the most optimal model, with accuracy reaching 88%. In addition to providing high accuracy, this model also shows good stability compared to the RGB and GLCM models alone. This proves that the use of a combination of features is more effective and reliable in identifying the marketability of oil palm FFB compared to the use of a single feature.]]>
Copyrights © 2025
