Technology and innovation

Research Undertaken by Post-doctoral Researcher Dr Enrique Valero, Dr Frédéric Bosché and Dr Alan Forster 2015-2022 

This research project focused on creating accurate and cost-effective digital building fabric inspections for conservation and maintenance, using terrestrial laser scanning data1.  

The project developed innovative computer vision methods for the processing of digital 3D data of the masonry to identify the individual stones and the mortar regions, both in ashlar and rubble masonry structures. Mortar recess - the difference (in depth) between the face of the stones and the face of the mortar joints between them - is also estimated. This allows rapid and accurate quantification of, for example, the number of stones in a façade and the quantity of mortar required for re-pointing (see image). Automatic statistical review of stone sizes and shapes also supported the study of phases of construction and interpretation of building histories.  

We have made the rubble masonry segmentation algorithm2 developed freely available to the heritage and construction sector. These algorithms have significant practical value for improving accuracy, speed and objectivity of historic facade inspections. A plugin has been developed for the widely-used open-source software package, CloudCompare, to allow automatic and semi-automatic segmentation of rubble masonry facades and output results in formats useful to, for example, architectural technicians and building surveyors.  

While this research is driving forward our conservation and maintenance within the properties we look after, it has the potential to fundamentally transform conservation practice within the wider heritage sector. Find out more in our recent publications and read our press release to celebrate the launch of the Cloud Compare plugin:  

• Automatic segmentation of 3D point clouds of rubble masonry walls, and its application to building surveying, repair and maintenance
• High Level-of-Detail BIM and Machine Learning for Automated Masonry Wall Defect Surveying
• Github Cyberbuildlab
• New innovative technology launched to help care for Scotland’s traditional buildings

Research Undertaken by Post-doctoral Researcher Dr Enrique Valero, Dr Frédéric Bosché and Dr Alan Forster 2018-2022 

This research project contributes to the collaboration’s aim to create accurate and cost-effective digital building fabric inspections for conservation and maintenance, using terrestrial laser scanning and photogrammetric data 3at its core. 

The project used machine learning for the interrogation of digital 3D data for historic building fabric condition survey to evaluate areas of deterioration.  Innovative algorithms were developed to detect and classify stone wall defects, in coloured 3D data of individual stones (obtained thanks to the segmentation tool described in Part 1). A machine learning -based model was developed that infer the presence of surface defects or manifestations of defects (e.g. erosion, delamination, cracks, discolouration) based on a range of geometric and colour based metrics. 

While this research is driving forward our conservation and maintenance within the properties we look after, it has the potential to fundamentally transform conservation practice within the wider heritage sector. Find out more in our recent publications: 

• Technical Paper 38
• Automated defect detection and classification in ashlar masonry walls using machine learning
• High Level-of-Detail BIM and Machine Learning for Automated Masonry Wall Defect Surveying

Research Undertaken by PhD Student Jiajun Li, 2022-2025 

Maintaining buildings and infrastructure is critical for sustainability of our built environment, the economy and the societal resilience of the region they service. English Heritage assert every £1 of invested in heritage-led regeneration has generated £1.60 in additional economic activity reflecting the underrepresented financial benefit of such interventions.  

Climate change projections for the UK suggest that the built environment in general and the historic/traditional built environment in particular, is being placed under increasingly significant strain, which raises fundamental challenges to the monitoring and maintenance of those structures. Such is the importance of human-caused climate change that in 2019, Scotland declared a climate emergency and, as a result, we published our Climate Action Plan that includes seven areas of action. The first area is ‘Climate Impacts and Adaptation’ and it is guided by a commitment to “continue to research and monitor the effects of climate change on the historic environment. The more we know, the better we can help it to adapt”. This PhD scholarship and project are set within this area of action. Within the context of ‘climate impacts and adaptation’ it is increasingly understood that external building fabric (roofs, walls etc) are particularly adversely affected by climate change and extreme weather events (storm, flood etc). 

This PhD scholarship was specifically directed to conduct a project that uses innovative data capture methods (e.g. drone surveying and photogrammetry) and information recording. It used multiple Artificial Intelligence (AI) data analysis methods to monitor slated roofs of traditional and historic buildings. The work first focused on the generation and semantic segmentation4 of roof orthoimages5 in order to obtain orthoimages of the slated areas only that are free of any occlusions (see figure). Then, various methods for slated roof defect detection were developed and compared, focusing on defects visible on the roof surface. The best solution was one implementing a two-stage approach with an initial slate detector followed by slate status classification (both employing some forms of Artificial Intelligence) to identify slates with organic growth or/and structural defects.   

This research is jointly funded by HES and an EPSRC Doctoral Training Partnership scholarship. 

• Occlusion-free Orthophoto Generation for Building Roofs Using UAV Photogrammetric Reconstruction and Digital Twin Data
• Extracting roof sub-components from orthophotos using deep-learning -based semantic segmentation
• Automated generation and semantic segmentation of roof orthophoto for digital twin -based monitoring of slated roofs
• Slate detection in orthophotos of building roof panels