Beyond the naked eye: Employing machine learning and computer vision to explore Iberian oculated idols decorative patterns

This is the authors' abstract. We don't add key points for this paper.

The study of the Copper Age in the Iberian Peninsula uncovers many enigmas, among them the oculated idols. Characterized by their distinctive eye-shaped motifs on long bones, these idols provide a window into the Chalcolithic communities' symbolic world. However, conventional archaeological techniques have fallen short in fully deciphering their narratives. This study uses advanced machine learning techniques to examine the decorative patterns on these oculated idols. We introduce a toolkit called ImageClusterVis, which utilizes computer vision algorithms to extract features more effectively. By applying Gaussian Mixture Models, Spectral Clustering, and Self-Organizing Maps, we can identify and classify patterns that are not visible to the human eye. Our approach helps uncover artistic variations and regional differences in the idols that have not been recognized in past research, showcasing the potential of machine learning to reveal new dimensions of ancient art. Lay Summary: The analysis of visual patterns in material artefacts has always been a central part of archaeological practice, but it has typically relied on qualitative methods. With the advent of digital archaeology, there has been a growing interest in developing more objective and quantitative techniques to compare and classify artefacts based on visual attributes. In this article, we present a novel methodology that employs computer vision algorithms and machine learning to extract robust visual descriptors and cluster archaeological images in an unsupervised manner. This study explores the world of Copper Age Iberian oculated idols on long bones by examining their decorative patterns, to uncover new insights into the regional distinctions, and cultural significances of these enigmatic figures, revealing a previously unseen depth to Copper Age society and artistry. This novel research not only advances our understanding of prehistoric Iberian cultures but also sets a new case study for the interdisciplinary application of AI in archaeology.