Toward a System for Post-Earthquake Safety Evaluation of Masonry Buildings

A quick and accurate post-earthquake safety assessment is critical for emergency management and reconstruction. Accurate knowledge of the scenario enables optimal use of human and economic resources. In Earth-quake prone countries, National Emergency Management Agency defines standard forms to collect information during inspections (e.g., Italian AeDES form, New Zealand Earthquake rapid assessment form, American ATC-20 Rapid evaluation safety assessment form). Assisting the technicians in the compilation of the cards and assessing their correctness guarantees a faithful reconstruction of the reality. We propose a Deep Learning-based tool that can recognize, localize, and quantify damages starting from a set of photos of the building to be assessed. The analysis results are expressed in terms of a Damage Assessment Matrix, which allows a quick association to the safety form.