Artificial intelligence for cultural heritage

Virtual restoration of stained glass windows

Collaborators

This is a collaboration with:

  • Isabelle Lecocq (art historian, KIK-IRPA)
  • A group of students from UMONS, alphabetically Lucas Colinet, Adrien Debuysschere, Aymeric De Volder, Roy Ebwelle, Rocherelle Fouefack, Oussama Hakik, Yvan Junior Tchuimadjio, Ariane Kouokam, Walid Mellouki, Gabriel Mroué, Franck Njampou, Louis Pluquet, Nathan Semet, Romane Sogalow. Supervised by Nicolas Gillis and Saïd Mahmoudi.

Results

Code

Open data

Description

Virtual restoration of stained glass windows addresses the following problem: given color photographs of a damaged window alongside grayscale images of the same window taken before the damage occurred, generate a colorized image representing the window as it would have appeared in its original state. This task combines image colorization and guided restoration, using the pre-damage grayscale as a structural reference and the surviving color regions as a color guide.

Virtual restoration of a stained glass window