Abstract

Determining the ripeness level of oil palm fruit is crucial in the harvesting process to ensure the quality and productivity of the yield. This study aims to evaluate the performance of the Convolutional Neural Network (CNN) algorithm in classifying oil palm fruit images into ripe and unripe categories. The dataset consists of images of fruits captured under various lighting conditions and angles. The CNN model achieved an overall accuracy of 91.75% on the test data. For the ripe fruit class, the model attained a precision of 97%, recall of 94%, and F1-score of 95%. Meanwhile, for the unripe fruit class, the model achieved a precision of 78%, recall of 87%, and F1-score of 83%. These results indicate that the CNN-based approach can be an efficient solution for improving the accuracy and efficiency of the oil palm fruit sorting process.