Lightning talk of dr. Martijn Kleppe at the launch of the new website of the KB Lab on 11 April 2017 on the KBK-1M dataset, which is available at http://lab.kb.nl/dataset/kbk-1m
3. How can we
extract a
caption of a
photo from a
website?
How can we find
images based on
the extracted
caption?
4. How can we create a tool that assists researchers
who want to research the re-use of photographs?
++ ++
==
5.
6.
7.
8. KBK-1M:
• 1,6 million images extracted from
digitised newspapers
• Photographs, cartoons, drawings
(‘Afbeelding met illustratie’)
• 1922-1994
• Combination of images & text
• Available for research purposes
• Possible use:
Humanities research questions
Computer Vision
Training dataset Deep Learning
9. Future Work
• Next researcher-in-residence:
Thomas Smits
• Expand dataset: 1860 – 1921
• Differentiate between engravings
and photographs
• Using Computer Vision
10. Any questions?
How to get access + 2 papers:
http://lab.kb.nl/dataset/kbk-1m
martijn.kleppe@kb.nl