Classification of Oil Palm Fresh Fruit Bunches (FFB) Based on Its Maturity Colour Using Convolutional Neural Network (CNN) Approach

Abstract

It is very crucial to identify the maturity of Fresh Fruit Bunches (FFB) of oil palm to ensure the appropriate timing is set for harvesting them and thus maintaining a good quality of oil palm production. In Malaysia, Oil Palm Industry is one of the biggest industries and has contributed a large amount of the country’s income. Therefore, it is important to check the maturity/ripeness of oil palm fruits before harvesting it. Currently, the grading of FFB based on its maturity is done manually by human and sometimes human can make mistake in performing the inspections. Hence, this paper proposes an automatic grading system of oil palm FFB using a deep learning approach, specifically by using Convolutional Neural Network (CNN). Most previously published works dealt with binary classification (Two classes classification). However, this work investigates the effectiveness of applying a deep learning approach to perform a multi-class classification task. In addition to that, not many works dealt with optimizing the deep learning model in classifying oil palm fresh fruit bunches (FFB) based on its maturity colour. Thus, this work will also evaluate the performance of the proposed CNN model with optimized training parameters that includes the number of epoch and batch size. Based on the results obtained, the number of epochs and the batch size have a great influence on the performance of the CNN models. For instance, having a high number of epoch’s value of 100 and bigger batch size’s value (e.g., 64), the CNN model can produce higher performance in term of accuracy.