Learning feature representation of Iberian ceramics with automatic classification models

摘要: Abstract In Cultural Heritage inquiries, a common requirement is to establish time-based trends between archaeological artifacts belonging different periods of given culture, enabling among other things determine chronological inferences with higher accuracy and precision. Among these, pottery vessels are significantly useful, their relative abundance in most sites. However, this very makes difficult complex an accurate representation, since no two these identical, therefore classification criteria must be justified applied. For purpose, we propose the use deep learning architectures extract automatically learned features without prior knowledge or engineered features. By means transfer learning, retrained Residual Neural Network binary image database Iberian wheel-made vessels’ profiles. These pertain sites located upper valley Guadalquivir River (Spain). The resulting model can provide feature representation space, which classify profile images, achieving mean 0.96 f -measure . This remarkably than state-of-the-art machine approaches, where several extraction techniques were applied together multiple classifier models. results novel strategies current research automatic objects study within Archaeology domain.