Every provenance researcher knows the feeling: you turn an artwork around and discover a mysterious label, stamp, handwritten note or inventory number. You know you've seen something similar before… but where?
Provenance marks are among the most valuable clues for reconstructing an artwork's history. Unfortunately, they are scattered across museum databases, publications, personal photo collections, and researchers' hard drives. Finding a match often means searching for hours—or simply hoping someone else has seen it before.
The goal is to develop a database that allows:
Users to upload and document provenance marks (labels, stamps, inscriptions, seals, inventory numbers, framing labels, transport labels, and more).
Each mark to be described using an existing controlled vocabulary developed by our provenance researchers.
Images and descriptions to be stored in a growing, searchable database.
Users to link entries to museum collection records or other online resources whenever possible.
Powerful filtering by mark type, location on the object, language, material, institution, or period.
As a dream feature: a reverse image search that helps answer the eternal question: "Have I seen this label somewhere before?"
As a bonus: a mobile app that allows researchers to photograph provenance marks directly in front of an artwork, search the database on the spot, and upload new discoveries in just a few clicks.
If all of this is a bit too ambitious for one hackathon, a prototype with image upload, structured descriptions, and a basic search based on descriptions would already make provenance researchers very happy (and probably save them quite a few hours).
For the development of the prototype, we will provide access to high-resolution images of the front and reverse of artworks via the IIIF interface of our digital collection platform, Sammlung Digital. In addition, participants will receive an existing Excel dataset containing numerous examples of object photographs, together with a curated selection of JPEG images to facilitate testing and development without the need to work with the entire collection.
The challenge owners have a background in provenance research and art history and already developed a structured nomenclature for describing provenance marks. We know the research problem—we just need your magic.
Missing skills in the team: Developers, AI & Computer Vision Enthusiasts, Database Wizards, UX Designers, GLAM Nerds, and anyone who enjoys solving little historical mysteries.
Please join us!
Keywords: Provenance Research, AI, Computer Vision, Image Matching, Reverse Image Search, Museum Collections, GLAM, Digital Humanities, Collaboration, Metadata, Mystery Solving.
Contact : Nora Rub, SKKG Winterthur
Data
Find a list of labels with provenance information (XLSX, CC0), a list with the metadata of 237 artworks with provenance information (CSV, CC0, incl. links to the IIIF manifests for the images and links to the object on https://digital.skkg.ch/), the provenance documentations for these works (PDF, [CC BY](https://creativecommons.org/licenses/by/4.0/ -> mention the provenance team and the year), as well as images of the front (CC BY -> provide the artist name and work information) and back (CC0) of the paintings: https://kdrive.infomaniak.com/app/share/136533/7b8ab950-c7c6-4647-9599-96540e2c087a/files/106358
The Public API gives access to “Sammlung digital”’s Linked Data Knowledge Graph.
-
Documentation: https://github.com/swiss-art-research-net/skkg-pipeline/wiki/6.-How-to-use-the-SKKG-Linked-Data
-
SPARQL endpoint: https://sparql.digital.skkg.ch/
-
or execute queries against the endpoint via QLever UI: https://qlever.digital.skkg.ch/
-
metadata and images of objects (CC0), images of artworks in the public domain (CC BY -> provide the artist name and work information), see SKKG’s Open Data Policy and Nutzungsbedingungen (Terms of Use).E914K6OQ.pdfE914K6OQ.pdf
Previous
GLAMhack 2026
Next project