Automatic damage detection of historic masonry buildings based on mobile deep learning

Abstract

Vision-based manual inspection technology for identifying and assessing superficial damage of historic buildings is time- and labor-consuming. To overcome these limits, this paper proposed a novel automatic damage detection technique using Faster R-CNN model based on ResNet101 framework to detect two categories of damage (efflorescence and spalling) for historic masonry structures. 33 different cases were studied, and the best case shown an average precision (AP) of 0.999 and 0.900 for efflorescence and spalling damage respectively, with a 0.950 mean AP. Moreover, an Internet Protocol (IP) webcam damage detection system combined with workstation was developed to detect the damage in real-time, and an automatic damage detection system based on smartphones was developed, which can realize real-time damage detection of brick masonry buildings. In addition, two on-site experiments were carried out on real masonry buildings to verify the feasibility and effectiveness of the system. Consequently, it was demonstrated that the proposed method was considerably automatic, efficient, and reliable for damage detection of historic masonry buildings and, ultimately, contributing to the management and protection of historic buildings.