4 files

Rijksmuseum Challenge 2014

posted on 03.12.2017, 10:30 by Thomas Mensink, J.C.vanGemert@tudelft.nl
The Rijksmuseum Challenge 2014 dataset is introduced in our ICMR 2014 paper "The Rijksmuseum Challenge: Museum-Centered Visual Recognition", T. Mensink and J. van Gemert (see details below). The challenge is to apply computer vision techniques on art objects (paintings, sculptures, drawings etc) from the Rijksmuseum (in Amsterdam, the Netherlands).

The dataset contains 4 files:
- rijksxml.tgz: raw xml files of about 100K objects (paintings, sculptures, drawings)
- rijksimg.tar: images belonging to the 100K objects
- rijksFV16.mat: Matlab file with Fisher Vectors (as used in our ICMR paper)
- rijksgt.mat: Matlab ground-truth file for the 4 challenges

Baseline classifiers/results
The Matlab code for running the four challenges, including our baseline classifiers based on the provided Fisher Vectors can be found at:

When using this dataset please cite the following paper

The Rijksmuseum Challenge: Museum-Centered Visual Recognition, Thomas Mensink and Jan van Gemert, ACM International Conference on Multimedia Retrieval (ICMR), 2014

author = {Thomas Mensink and Jan van Gemert},
title = {The Rijksmuseum Challenge: Museum-Centered Visual Recognition},
booktitle = {ACM International Conference on Multimedia Retrieval (ICMR)},
year = {2014}

See also the link to the pdf file (in the references)

License and copyright of the images:
The images are downloaded using the Public API of the Rijksmuseum. According to their conditions: "All data and all images made available through the API are either in the public domain or are subject to a CC0 license. The data and images are royalty-free and may be copied, distributed, modified and used without the permission of the Rijksmuseum."

More info on this copyright:

Retention Period
The retention of the dataset has been set to 10 years after publications (april 2024)


Retention period




University of Amsterdam / Amsterdam University of Applied Sciences