Videomining in historischem Material - ein Praxisbericht

摘要

Video mining algorithms such as concept classification and person recognition enable fine-grained semantic search in large video archives like the historical collection of the former German Democratic Republic (GDR) of the German Broadcasting Archive (DRA). We present the project VIVA, our deep learning approaches, and the VIVA software tool, which allows users to easily acquire data to train analysis algorithms.