Description

Over the history of human civilization, works of art and architecture have been subjected to fragmentation: ancient Mayan stele were cut away from monuments by collectors; medieval sculptures from Notre-Dame in Paris were broken into many parts and dispersed in acts of political iconoclasm. Art historians and archaeologists seek to reconstruct these works to more fully understand their cultural meaning and value, but the traditional method of physical refitting is labor intensive and not always possible when fragments are dispersed throughout the world. We propose to use AI in combination with existing digital scan models of fragments to develop a means for reconstructing fragmented cultural heritage artifacts in a virtual space.

Duration

Total project length: 175 hours

Task ideas

• Search for direct fit matches between surfaces (e.g. two parts of something broken)
• Identify continuity of carved topography (e.g. parts of the same carved feature, but with gaps).
• Identify continuity of surface designs (e.g. parts of the same carved feature, but with gaps).
• Identify broader dimensional resemblance (e.g. the shape of the stone blocks used to make that sculptural facade)

Expected results

• Develop machine learning models that can reconstruct digitized fragments with at least 80% accuracy
• Train AI model to search for matches between digitized fragments for which is original orientation is certain
• Test and train AI model using fragments for which original orientation is uncertain

Requirements

• Python and some previous experience in Machine Learning.
• Ability work with 3D scan data (e.g. .PLY and .OBJ files)

Project difficulty level

Medium

Links

• Notre Dame in color
• Visual Documentation Lab

Mentors

• Jennifer Feltman (University of Alabama)
• Alexandre Tokovinine (University of Alabama)
• Emanuele Usai (University of Alabama)
• Sergei Gleyzer (University of Alabama)

Please DO NOT contact mentors directly by email. Instead, please email human-ai@cern.ch with Project Title and include your CV and test results. The mentors will then get in touch with you.

Corresponding Project

• HealingStones

Participating Organizations

• Alabama

---