Skip to main content
Have a personal or library account? Click to login
What to do with 2.000.000 Historical Press Photos? The Challenges and Opportunities of Applying a Scene Detection Algorithm to a Digitised Press Photo Collection Cover

What to do with 2.000.000 Historical Press Photos? The Challenges and Opportunities of Applying a Scene Detection Algorithm to a Digitised Press Photo Collection

Open Access
|Jun 2022

Figures & Tables

Figure 1

Ria Kuyken attacked by a circus bear (1962). Photo taken from De Boer Collection (collection Fotopersbureau De Boer, 01806A00_08, Noord-Hollands Archief).

Figure 2

The logs of four days in 2001. In this period, we see many photographed events per day and a strong regional focus with almost all photos sold to the local newspapers Haarlems Dagblad and IJmuider Courant. Picture taken by Paul Maessen (Noord-Hollands Archief).

Figure 3

One side of the subject index card ‘first post’. This refers to pictures of the first post in the ground construction, illustrating that the agency did not just take photographs of the unique and extraordinary but also of the mundane.

Table 1

Top 10 topics (out of ca. 1.500) of the subject index cards of Fotopersbureau De Boer (1945-1990), with the number of cards.

SubjectNumber of cards
Persons 141
Soccer 74
Buildings: exterior 64
Portraits 60
Openings, commissioning 59
Schools 49
Music 40
Fire, fire brigade 36
Buildings: interior 35
Construction works 35
Figure 4

Scene Prediction. The model correctly classified the picture as ‘Dining Room’. The predicted labels ‘portrait_group’ and ‘marriage’ also apply.

Figure 5

The distribution of scene categories that appeared more than fifty times in the pilot training set. We used this annotated set to re-train the places-365 model.

DOI: https://doi.org/10.18146/tmg.815 | Journal eISSN: 2213-7653
Language: English
Page range: 1 - 24
Published on: Jun 7, 2022
Published by: Sound & Vision
In partnership with: Paradigm Publishing Services

© 2022 Melvin Wevers, Nico Vriend, Alexander de Bruin, published by Sound & Vision
This work is licensed under the Creative Commons Attribution-ShareAlike 4.0 License.