Deep learning for detecting distresses in buildings and pavements: a critical gap analysis

Purpose The massive number of pavements and buildings coupled with the limited inspection resources, both monetary human, to detect distresses recommend maintenance actions lead rapid deterioration, decreased service life, lower level increased community disruption. Therefore, this paper aims at providing a state-of-the-art review literature respect deep learning techniques for detecting distress in buildings; research advancements per asset/structure type; future recommendations applications detection. Design/methodology/approach A critical analysis was conducted on 181 papers learning-based cracks structured adopted so that major articles were analyzed according their focus study, used methods, findings limitations. Findings utilization pavement is advanced compared assess evaluate structural health buildings. There need studies compare different convolutional neural network models foster development an integrated solution considers data collection method. Further required examine setup, implementation running costs, frequency capturing tool. In conclusion, applying algorithms lieu manual has shown promising results. Practical implications availability previous improvements proposed computational tools (e.g. artificial intelligence, learning, etc.) are triggering researchers practitioners enhance distresses’ process make better use resources. Originality/value crack detection first time enable novice highlight knowledge gap each article, as well building base from other support developing workable solutions.