I-MEDIA-CITIES: Automatic Metadata Enrichment of Historic Media Content

Abstract

Cultural, heritage and museum institutions across Europe are increasingly holding large digital collections, either digitized or born digital, that can have a significant impact on many research fields, both in Social Sciences and Humanities. This being a somehow recent development, a strong, efficient link between these institutions and the research field still largely needs to be established, as well as a true bi-directional dialogue that aligns the efforts and the focus of these institutions to the needs of research. Equally important is the fact that neither cultural, heritage institutions nor the research field are taking full advantage of information technologies in order to improve the opportunities and the quality for research and for providing access to these collections. These two factors result in a general lack of significant research and experience in intelligent, technology-intensive, and user-engaging solutions designed to foster research on cultural digital collections across Europe.In this paper we adapt and apply a variety of recently developed techniques in the field of computer vision and machine learning in order to enrich historic media content with automatically generated metadata for faster and more efficient search in digital collections and archives. We particularly focus on moving images, especially high quality media content related to the history, sociology and culture of European Cities from the 19th to the 21st century.