Automated engineering analysis of crack mechanisms on building façades using UAVs

Cracking on building façades potentially poses significant safety hazards. However, traditional manual inspection of these facades is known to be laborious, time-consuming, costly and dangerous. Unmanned Aerial Vehicles (UAVs) and computer vision techniques are increasingly used for crack condition evaluation on building façades but typically fall short of conducting engineering analysis. To address this limitation, it is necessary to automatically interpret cracking mechanisms, i.e. whether structural or non-structural. This paper introduces a vision-based framework for UAV-assisted building façade crack mechanism assessment. In the framework, the contextual information is identified and extracted to support engineering analyses of cracks on building façade, and a rule-based approach for crack mechanism assessment by leveraging extracted crack characteristics and contextual information is formalized. The effectiveness of the framework was evaluated through real-world applications on two high-rise buildings. The experimental results demonstrated that the framework successfully identified both structural and non-structural cracks with precision and recall rates over 80 %. • Vision-based framework for façade crack mechanism assessment using UAV is proposed. • Contextual information of concrete cracks is extracted for engineering analyses. • Rule-based approach is developed to distinguish structural and non-structural cracks. • Validation on realistic buildings positions it as a practical tool for UAV-based crack assessment. • Results show precision and recall rates of over 80 % on real-world applications.