Learning Feature Representation of Iberian Ceramics with Automatic Classification Models

Abstract

In Cultural Heritage inquiries, a common requirement is to establish time-based trends between archaeological artifacts belonging to different periods of a given culture, enabling among other things to determine chronological inferences with higher precision. Among these artifacts, pottery vessels are significantly useful, given their relative abundance in most archaeological sites. However, this very abundance makes difficult and complex an accurate representation, since no two of these artifacts are identical, and therefore classification criteria must be justified and applied. For this purpose, in this work we propose the use of deep learning architectures to extract automatically learnedfeatures without prior knowledge or engineered features. By means of transfer learning, a Residual Neural Network was retrained with a binary image database of Iberian wheel-made pottery vessels’ profiles. These vessels pertain to archaeological sites located in the upper valley of the Guadalquivir River (Spain). The resulting model is able to provide an accurate feature representation space, which is able to classify profile images automatically, achieving a mean accuracy of 0.98. This accuracy is remarkably higher as compared with other state of the art machine learning approaches, where several feature extraction techniques were applied together with multiple classifier models. Furthermore, we show the relevance of introspection in our automatic feature extraction method prior to classification, and the effects of poor feature selection. These results provide novel approaches to current research in automatic feature representation and classification of different objects of study within the Archaeology domain.