Klasifikasi Citra Motif Batik Menggunakan Convolutional Neural Network Ensemble

Indonesia is one of the countries in the world that possesses rich cultural wealth in each of its regions. One of the enduring cultural treasures in Indonesia is Batik. The abundance of Batik motifs in Indonesia requires individuals to invest time and specialized knowledge in manually identifying Batik to accurately describe and classify these distinctive patterns. This research proposes the classification of Batik motif images using the CNN Ensemble method for identification of patterns in East Javanese Batik. Two scenarios will be applied to the dataset for identification, along with the implementation of the sliding window preprocessing technique. In this study, two scenarios are executed. In Scenario 1, the CNN Ensemble model without augmentation processes yielded an accuracy of 95% and a loss of 0.232. Meanwhile, in Scenario 2, the CNN Ensemble model with augmentation processes resulted in an accuracy of 94% and a loss of 0.201. In this research, the augmentation technique did not significantly improve the accuracy
values provided by the CNN Ensemble model.