Scene Detection in De Boer Historical Photo Collection

Open Access
Authors
Publication date 2021
Host editors
  • A.P. Rocha
  • L. Steels
  • J. van den Herik
Book title ICAART 2021 : Proceedings of the 13th International Conference on Agents and Artificial Intelligence : February 4-6, 2021. - Volume 1
Book subtitle ARTIDIGH 2021, Vienna, Austria
ISBN
  • 9789897584848
Event Special Session on Artificial Intelligence and Digital Heritage
Pages (from-to) 601-610
Number of pages 10
Publisher Setúbal: SciTePress
Organisations
  • Faculty of Humanities (FGw) - Amsterdam Institute for Humanities Research (AIHR) - Amsterdam School of Historical Studies (ASH)
Abstract This paper demonstrates how transfer learning can be used to improve scene detection applied to a historical press photo collection. After applying transfer learning to a pre-trained Places-365 ResNet-50 model, we achieve a Top-1 accuracy of .68 and a Top-5 accuracy of .89 on our data set, which consists of 132 categories. In addition to describing our annotation and training strategy, we also reflect on the use of transfer learning and the evaluation of computer vision models for heritage institutes.
Document type Conference contribution
Language English
Related dataset Fotopersbureau De Boer Training Set on Scene Detection HisVis: Scene Detection Pilot Training Set Fotopersbureau De Boer Training Set on Scene Detection
Published at https://doi.org/10.5220/0010288206010610
Downloads
ARTIDIGH_2021_1 (Final published version)
Permalink to this page
Back