All ancient architectural features, monuments, and smaller artifacts inevitably break. The breaks may be accidental or deliberate, stemming from ritual practices, political iconoclasm, or actions of modern looters and collectors. Ancient Maya would ‘de-activate’ building facades, carved monuments, and ceramic vessels by breaking and caching them in multiple locations. Other Maya monuments were broken as acts of political and ritual violence. Political and religious upheaval was responsible for breaking and dispersing of European medieval sculptures including those of the Notre Dame cathedral in Paris. Identifying and matching broken pieces is often an essential and highly labor-intensive component of research and conservation projects in Archaeology and Art History. The challenge becomes especially daunting when fragments of the same sculpture end up dispersed across various private and institutional repositories. We believe that new methods of digital documentation combined with machine learning can greatly enhance our ability to reconstruct ancient monuments and artifacts.
This is the second phase of the project. It builds on the outcomes of the first phase that identified several promising approaches to identifying and refitting matching sculpture fragments. At the same time, the first phase highlighted several challenging tasks such as finding matches between segments of vastly different sizes, classifying fragments surfaces, and dealing with match imperfections caused by documentation errors and surface erosion. We hope to resolve these issues while working with the same test dataset: fragments of a single ancient Maya monument that was deliberately broken and cut by modern looters. The dataset offers a wide range of match types, from closely fitting breaks to fragments sharing surface topology yet separated by gaps.
Total project length: 175 hours
Medium
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.