4TU.ResearchData dataset wins the Dutch Data Prize 2022

The 4TU.ResearchData dataset from Mitchell van Zuijlen’s PhD project called “Materials in Paintings (MIP): An interdisciplinary dataset for perception, art history, and computer vision” won the Dutch Data Prize in the natural & engineering sciences category this November 2022. The prize was collected during the awards ceremony by our colleague Jan van der Heul, who personally curated the dataset. 

“I am so happy to see that the datasets I have curated won the Dutch Data Prize this year! Makes me so proud that the curation work we do in 4TU.ResearchData to improve the FAIRness of datasets really matters”, says Jan.

MIP is an annotated dataset of 19,000 paintings from the past 500 years! Its reusability across different domains and its applicability to the arts and humanities make the dataset very special.

The paintings were cut into more than 200,000 fragments. The depicted materials in each of these fragments were classified using machine learning algorithms. Of course, the cutting was done on digital images – and all can be downloaded in open formats, both as a comprehensive dataset available via 4TU.ResearchData and via an interactive portal that allows users to browse through the individual paintings. 

The dataset got downloaded over 1200 times in 28 countries only 1.5 years after uploading it. “When I was presenting my research at conferences, people were often positive about the idea of publishing the data open access,” Mitchell says. “However, you never know what will happen when you publish the data. It was a big relief to finally put it online and very cool to see all these people using it now.”

Mitchell also believes that the open science community, including the 4TU.ResearchData and its support, is becoming more and more reliable and will save new researchers a lot of time and effort in the future. Mitchell conducted this work as part of his PhD project at the Delft University of Technology, Faculty of Industrial Design Engineering. A big congratulations for the Dutch Data Prize win!

Links to Mitchell’s data:

• Dataset: Materials In Paintings (MIP): An interdisciplinary dataset for perception, art history, and computer vision (https://doi.org/10.4121/13679200.v1)
• Website: Materials in Paintings Dataset (https://materialsinpaintings.tudelft.nl/)