AI Supported Automated Construction Compliance Inspection

Construction inspection is critical to project delivery in the Built Environment, yet it remains largely manual on construction sites. In person inspections have been found to have error rates of 20-30%. Further, inspections are often performed on a sampling basis, examining only a fraction of the work. Although this saves time, but it limits coverage and increases the risk of accidents. As a result, these inefficiencies lead to quality issues, cost overruns and schedule delays.

To address this problem, our collaborative research project funded by Construct Innovate (Grant No. CISFC1-24_012) is aimed at developing a ‘human-in-the-loop’ automated construction inspection solution through machine learning. This project intends to leverage mobile cameras and deep learning computer vision to identify potential non-compliance issues at critical wall junctions. The project is supported by our industry partner Fenagh Engineering and Testing.