<![CDATA[Topeng Pajegan is a Balinese cultural heritage distinguished by specific visual carving patterns and holds significant aesthetic, philosophical, and religious value. However, its classification is still conducted manually and depends on individual expertise, making it prone to subjectivity and limited scalability, while digital documentation remains limited. This study proposes a classification system for Balinese Topeng Pajegan using a Convolutional Neural Network (CNN) to support cultural digitalization and preservation. The system was developed following the CRISP-DM methodology, utilizing a dataset from Kaggle complemented by authentic data collected from Balinese mask artisans, dancers, and collectors. Model optimization was performed through data augmentation and RandomSearch-based hyperparameter tuning. Experimental results demonstrate that the optimized CNN model successfully classified six types of Topeng Pajegan with an accuracy of 90.12%, supported by F1-score and confusion matrix evaluations. The model was implemented in a web-based application, which functions as both an automated classification tool and a digital educational platform to promote the sustainable preservation of Balinese cultural heritage. Keywords: Topeng pajegan, Convolutional Neural Network, Computer Vision]]>
Copyrights © 2026
