An Application of Convolutional Neural Networks for Archaeological Andean Pottery
Dina Soledad Cornejo Meza, Mauricio Salazar Espinosa, Javier Vera Zúñiga · 2024
This paper presents an initial exploration into the application of Convolutional Neural Networks to archaeological Andean region pottery. To do this, it studies how Artificial Intel-ligence can make cultural artifacts more accessible and understandable to a wider audience, thereby democratizing knowledge. Utilizing a comprehensive database from the Larco Museum, which includes 44,711 archaeological artifacts, this study focuses on a refined subset of 29,115 entries. We enhance this dataset with additional archaeological annotations and employ data augmentation techniques to train the Convolutional Neural Networks models. The paper demonstrates significant results in accurately predicting the style, culture, and chronology of these artifacts, showcasing the potential of Artificial Intelligence to help archaeological research. The paper underscores the transfor-mative impact of Artificial Intelligence integration into Andean archaeology, not only as a methodological enhancement but also as a tool for better preservation and understanding of cultural heritage. Looking ahead, we propose further improvements to the models by expanding the dataset and refining annotations to better support the cataloging processes in museums and other cultural institutions.