The dataset comprises of imagery/photographs collected using a standard DSLR camera, from first hand site visits and inspections. These images consist of various damage classes such as crack spalling and intact areas which were collected from three national heritage sites in the Western Cape namely: the Castle of Good Hope, Dal Josafat and Robben Island.
The approach used a quantitative research methodology, utillising machine learning and deep models for the analysis and classification of the image dataset into the various damage classes.