Algorithmic Representation of Batik Motifs: Visual Classification as a Form of Digital Èkphrasis

In recent years, artificial intelligence has assumed an increasingly prominent role across various re-search fields, extending its scope to include the representation and study of architecture and cultural heritage. This rapidly evolving domain highlights the potential of AI technologies to provide innovative tools for the analysis, interpretation, and understanding of visual and decorative systems embedded in artistic and architectural traditions.This paper examines the application of CNNs (Convolutional Neural Networks), with particular focus on the ResNet50 model, for the automatic classification of Batik patterns –a traditional form of visual expression with profound symbolic and cultural significance. Through the analysis of a dedi-cated dataset, the model training process, and the interpretation of results, the study investigates the potential of algorithmic description as a new form of èkphrasis –understood as a practice of transla-tion between visual and textual languages, here mediated by computational methods.The proposed approach integrates computer vision techniques with theoretical reflections on the relationship between digital representation and visual culture, emphasizing how artificial intelligence can contribute not only to the automation of classification processes but also to the understanding, enhancement, and preservation of artistic and decorative heritage.