Metal-detecting is a hugely popular pastime in the UK. Every year, thousands of enthusiasts regularly discover coins, dress accessories and other artefacts – representing millennia of British history – in ploughed fields across the country. These finds, which would otherwise remain unknown to archaeology, constitute a uniquely informative, ever-growing corpus which has transformed our understanding of certain periods.
The Portable Antiquities Scheme (PAS), based at the British Museum, allows detectorists across England and Wales to report their finds, so that they can be recorded in a central, curated, open-access database. The initiative has been so successful that the database now contains more than 1.6 million items, embracing a plethora of artefacts of different types. But the scale of this success presents new challenges: data entry by a wide range of individuals can be time-consuming and prone to inconsistency or error, while conventional research methods struggle to grapple with the sheer abundance and diversity of the data.
The AntiquAI project brings machine learning to bear on this ‘big data’ archaeological challenge, applying cutting-edge computer vision systems to classify PAS finds based on their photographs. In this initial twelve-month pilot project, we will establish how far these artefact photographs are conducive to image recognition, and so demonstrate the potential for AI to support assistive classification and analysis tools for users of PAS data, including enthusiasts, volunteers, heritage professionals, and researchers.
For more information, please contact mark.mckerracher@bodleian.ox.ac.uk
