Classification of Batik Cual Bangka Belitung Based on Deep Learning: YOLOv11 Approach

Abstract

The rapid growth of artificial intelligence, particularly deep learning, has enabled significant advancements in computer vision and automated image recognition. However, the application of these technologies to traditional cultural artifacts remains limited, especially within the domain of Indonesian textile heritage. Batik Cual Bangka Belitung, which features intricate ornamentation and visually similar motifs, presents unique classification challenges that conventional Convolutional Neural Network (CNN) models struggle to address effectively. To overcome these limitations, this study introduces an automatic motif classification system using the YOLOv11 architecture, a state-of-the-art object detection model capable of identifying and distinguishing fine-grained visual patterns. The research follows a systematic pipeline consisting of dataset collection, curation, manual motif labeling, image preprocessing, model configuration, training, and testing. A curated dataset of Batik Cual images was augmented and divided into training, validation, and testing subsets to ensure robust evaluation. Experimental results demonstrate strong model performance, achieving a precision of 0.934, recall of 0.808, mAP50 of 0.950, and mAP50–95 of 0.8172. These findings confirm that YOLOv11 can accurately detect motif regions and classify them under varying structural and textural conditions. The study contributes not only a reliable technical framework for recognizing Batik Cual motifs, but also supports digital preservation efforts and future cultural computing applications.