What challenge or opportunity did your project address, and what activities, methods, or outcomes have been achieved so far?
The project addresses one of the major challenges in cultural heritage digitisation: reconstructing heritage artifacts when only a single historical photograph or illustration is available. Traditional methods such as photogrammetry and laser scanning require direct access to the object and extensive image datasets, conditions that are often impossible for lost, inaccessible, or endangered heritage.
To overcome these limitations, I developed an AI-based workflow for automated 3D reconstruction from a single image using open-source deep learning models, combined with post-processing and expert validation. The methodology was tested on different case studies: inaccessible items in museum, or difficult to survey, or heritage lost or destroyed. The resulting models proved suitable for morphological comparison, metric analysis, and the study of gameplay hypotheses, demonstrating that AI can significantly reduce the time and manual effort required for digital reconstruction.
What key insights, lessons learned, or innovations emerged from your work that could be valuable to others working in the fields of cultural heritage and 3D technologies?
The main innovation of the project is the automation of a workflow that has traditionally relied on manual 3D modelling. AI-based reconstruction enables faster, reproducible, and scalable generation of digital models from extremely limited visual evidence, making it particularly valuable in emergency documentation and heritage-at-risk scenarios. The research also highlights the importance of combining computational methods with expert interpretation. While AI can generate convincing geometries, the reliability of the results depends on image quality, algorithm training, and transparent validation. The project therefore demonstrates that AI should complement domain expertise, providing a practical framework that balances automation with scientific rigor.
How could your project’s results, expertise, tools, datasets, or approaches contribute to and benefit the broader Time Machine Organisation network, including opportunities for collaboration, knowledge exchange, or reuse by other members?
The proposed workflow is fully reproducible using open-source software and can be adapted to a wide range of cultural heritage contexts where limited documentation is available. It offers a practical solution for creating 3D assets from historical photographs, archival material, or isolated images, complementing existing digitisation strategies within the Time Machine ecosystem. The methodology could support collaborative initiatives on endangered heritage, museum collections, and archaeological archives by providing an efficient approach to generating reusable 3D content. I would welcome collaboration with other TMO members to benchmark AI reconstruction methods, integrate semantic metadata, develop shared datasets for heritage-specific AI training, and explore applications in virtual museums, digital twins, immersive experiences, and experimental archaeology. Ultimately, the project aims to contribute to a more accessible, scalable, and interdisciplinary approach to cultural heritage digitisation.
A short personal statement reflecting on my experience with the Academy
Participating in the Academy has been an inspiring and valuable experience. It provided the opportunity to engage with researchers and professionals from different disciplines, broaden my perspective on digital cultural heritage, and receive constructive feedback on my research. The interdisciplinary environment encouraged new ideas and collaborations, reinforcing my interest in applying artificial intelligence and 3D technologies to cultural heritage. The Academy has also strengthened my motivation to contribute to the Time Machine community by sharing methodologies, learning from others, and developing innovative approaches to heritage digitisation.