Klasifikasi Kematangan Buah Kelapa Sawit Menggunakan Metode Faster Region Convolutional Neural Network (Faster R-CNN)

Abstract

Indonesia is one of the world's largest producers of palm oil. Crude Palm Oil (CPO) is one of the raw materials with high utility, both for daily needs and industrial purposes. The quality of CPO produced depends greatly on the maturity level of the oil palm fruit. The determination of the fruit's maturity level is generally based on the number of fruit bunches and their color, making the harvesting of oil palm fruit an important activity in improving the quality of CPO (Fauzi, 2007). Unripe oil palm fruit is typically black in color, partially ripe fruit is reddish-black, and fully ripe oil palm fruit appears red to orange in color. Manual identification of maturity levels is time-consuming and prone to errors, leading to inaccurate determinations by farmers. A sustainable method is needed to identify oil palm fruit with good maturity levels to maintain CPO quality and increase crop yields. In this study, the Faster Region Convolutional Neural Network (Faster R-CNN) method was used to classify oil palm fruit into three maturity levels: ripe, underripe, and unripe. A total of 3,600 data points were used, consisting of 2,880 for training, 360 for validation, and 360 for testing. The application of Faster R-CNN in classifying the maturity levels of oil palm fruit resulted in an accuracy rate of 95%.