Machine learning based approaches for the visual analysis of historical artefacts
Dr R Evans
Dr K Rodriguez-Echavarria
No more applications being accepted
Funded PhD Project (European/UK Students Only)
The pace at which content for Arts, Heritage and Archaeology is being acquired continues to accelerate in both the raw volume acquired and the variety of datatypes being recorded. For instance, Fraunhofer-IGD has developed technology which automates the acquisition of 3D digital models of cultural artefacts, making it possible for museums and other cultural organisations to digitise their collections on a large scale. However, building large collections of 3D models, and other digital assets (such as text, images, video, manuscripts etc.), brings with it new problems of search, access and presentation. Such processes can be supported by organising and classifying content together and providing searchable representations of properties such as shape, material or style (for example). This project aims to develop visual analysis approaches, based on machine learning methods, to organise and classify 3D models which can be integrated into cultural heritage practice to enhance the management, accessibility and experience of digitised cultural heritage.
The key research questions to be addressed are:
1. What challenges does the advent of large scale 3d digitisation bring to the organisation and management of museum collections?
2. What new opportunities for visual analysis, based on machine learning methods, arise from the availability of large scale digitised collections?
3. Can the provision of scalable acquisition tools have a significant impact on cultural heritage practice, workflows and standards in an age of large-scale digitisation?
The research methodology will be to:
1. Develop research scenarios in conjunction with the cultural collaborator to hypothesise new approaches for the visual analysis of 3D models that could be empowered by the availability of large-scale digital asset collections.
2. Select an experimental set of artefacts to explore the practicalities of machine learning techniques for organising and classifying large scale digital collections.
3. Engage with potential CH researchers to develop web-based 3D-centered tools to support the visual analysis of digital collections.
4. Evaluate the degree to which the organisation and linking of digital collections’ metadata can be effectively automated and produce results that would not have been anticipated without the use of the technologies
The University of Brighton will provide expertise in knowledge engineering, cultural heritage ontologies and digitisation campaigns. Fraunhofer-IGD will provide expertise in and datasets resulting from mass digitisation and access to/involvement in a re-engineered 3D-repository infrastructure. The cultural partners will provide access to original artefacts and expertise in the nature of collections and the types of knowledge that it is desirable to detect within them.
This project is a four year integrated MRes/PhD studentship in the SEAHA Centre for Doctoral Training (www.seaha-cdt.ac.uk), funded by the EPSRC, UCL, the University of Brighton, the University of Oxford, and SEAHA heritage and industrial partners. The project is fully funded for UK resident students, and fees-only for other EU students. Limited possibility of funding for international students may also be available.