Classification of Oil Palm Fruit Ripeness Using Artificial Neural Network

Oil extraction rate (OER) and quality of oil palm can be improved by precisely and accurately classifying the ripeness of oil palm before harvesting. This paper focuses on the development of an artificial neural network (ANN) model for classification of oil palm fruit ripeness using Raman spectra features. In this study, the oil palm fruitlets are from the dura x pisifera (DxP) progenies, harvested from the oil palm plantation of National University of Malaysia (UKM) managed by Khazanah-UKM. A total of 50 samples from unripe, over ripe and ripe fruitlets were collected according to the standard of Malaysia Palm Oil Board (MPOB). Raman spectra for each sample are collected from benchtop Confocal Raman spectrometer. The spectral features for each sample are extracted using pre-processing techniques and used as predictors to train the ANN model. Samples are divided into training set and test set using 50:50 holdout method. The developed model achieves 95.48% prediction accuracy. The accuracy and robustness of the neural network can be improved by increasing the number of samples used in the training.