The “Visual Completion of Incomplete Historical Artifacts Using Artificial Intelligence” project presents an environmentally responsible and innovative leadership approach to cultural heritage restoration.
The project aims to digitally reconstruct missing parts of damaged historical artifacts using deep learning techniques before physical restoration efforts are undertaken.
This “digital-first” approach provides a model that significantly reduces environmental impact by minimizing the use of traditional restoration resources, chemicals, and materials.
The project aims to digitally reconstruct missing parts of damaged historical artifacts using deep learning techniques before physical restoration efforts are undertaken.
Traditional methods involve resource-intensive processes such as cleaning with specialized solvents and reproducing rare and costly ancient pigments.
The project aims to eliminate waste and trial-and-error steps through virtual simulations.
Restoration options can be evaluated before any physical intervention is carried out.
Resources can be allocated only to the most necessary and effective physical restoration efforts.
Increase in the SSIM score in a study applying a model architecture such as U-Net.
Reduction in model size achieved in the same study.
Recovery rate of original information from stone reliefs using a single 134-year-old photograph in the Borobudur Temple project.
The project goes beyond a single restoration effort by presenting a transferable leadership model for the cultural heritage sector.
In addition to focusing on technological implementation, the project plan proposes strategic solutions to sector-wide challenges such as data scarcity, algorithmic bias, and ethical concerns.
One of the key recommendations of the project is to establish multidisciplinary teams consisting of artificial intelligence experts, archaeologists, art historians, ethicists, and legal professionals.
This approach ensures not only the effective development and implementation of the technology, but also the ethical evaluation of its results.
The project contributes to the development of international standards by promoting the digitization, standardization, and inter-institutional sharing of cultural heritage data.
This enables algorithms to be trained on more diverse and less biased datasets, while creating a collaborative infrastructure across the sector.