Deteksi Kematanagan Buah Sawit dengan Menggunakan Algoritma Convolutional Neural Network

Abstract

This research aims to develop an automatic palm fruit ripeness detection system using the Convolutional Neural Network (CNN) algorithm. The dataset used consists of thousands of images of ripe and unripe palm fruits with varying lighting conditions and shooting angles. The CNN model used is MobileNetV2 which has been adapted for binary classification tasks. The training process is performed using data augmentation techniques to improve the generalization of the model. The evaluation results show that the developed CNN model is able to classify the ripeness of palm fruits with an accuracy of 84%. Comparison with conventional methods that rely on visual assessment shows that the CNN model provides more consistent and objective results. The implementation of this model has the potential to increase the efficiency of the harvesting and processing of palm fruits and reduce production costs.