Crack Detection and Monitoring: Review and Comparison of IoT and Image-Based Methods [Roadmap for Measurement and Applications]

The aging of critical civil infrastructure, such as bridges and buildings, and their vulnerability to damage often requires inspection and predictive maintenance tasks. Physical damage in the structure can represent danger and potentially lead to catastrophic consequences, such as the collapse of the structure. Cracks usually emerge as small fissures in the surface of infrastructure components, making them weak and potentially leading to structural failures. The early detection of such cracks and their continuous monitoring leads to prompt intervention and increases the safety and lifetime of the monitored infrastructure [1]. For this reason, several studies proposed automatic techniques to detect and monitor cracks by analyzing their length, width, depth, and severity [2]–[4]. The advent of Machine Learning (ML), Deep Learning (DL), and computer vision techniques created a new breed of image-based methods to detect cracks and monitor them over time [5]. However, ML/DL methods require a preliminary phase of model training, in which a crack image dataset must be collected and labeled. Moreover, risks significantly limit human accessibility for building inspections, particularly in external areas. Therefore, numerous researchers propose the integration of unmanned aerial vehicles (UAVs) to perform autonomous image collection from a target structure [6]–[8]. Although image-based methods are widely used for detection purposes [4], [5], they have many limitations. Areas that are always obscured prevent the utilization of those techniques. Further, detecting and monitoring cracks through images is challenging in non-uniform structures, such as masonry and cultural heritage (CH) buildings, as well as in large structures due to the extensive surface area needed to analyze. In those scenarios, the joint employment of sensor measurements with image-based techniques has the potential to enhance the accuracy and precision of crack monitoring systems [2].