Maturity Grading of Oil Palm Fresh Fruit Bunches Based on a Machine Learning Approach
Anindita Septiarini, Heliza Rahmania Hatta, Hamdani Hamdani, Ana Oktavia, Anita Ahmad Kasim, Suyanto Suyanto
Abstract
Grading maturity oil palm fresh fruit bunches (FFB) is an essential issue in the agriculture sector because the quality of palm oil determines based on the maturity level. Recently, the production of high-quality palm oil has increased continually. Therefore, the implementation of computer vision in agriculture for grading the maturity of oil palm FFB is required to avoid subjectivity in determining the maturity level. This study develops a classification method for grading the maturity level of FFB. Generally, this classification method performed using the color feature. In this study, the color feature is used to distinguish the maturity level of oil palm FFB. The mean value extracted as the color features in L*a*b color space is followed by implementing a machine learning method: Linear Discriminant Analysis (LDA), in the classification process. The experiment used a dataset of 150 images with three different classes: raw, under-ripe, and ripe. The dataset is applied in two stages: training and testing of 60 images and 90 images, respectively. The performance evaluation of the method used successfully achieved an accuracy value of 98.89% using a testing dataset of 90 images.