LIDAR, DRONES AND BRIM FOR RAPID BRIDGE INSPECTION AND MANAGEMENT

Abstract

Cracks are one of the main defects on concrete surfaces, and they are indicators of concrete structures’ state of health. Because traditional methods to identify and assess cracks rely on manual measurements, a significant number of studies have focused on identifying ways to automate this process. Accordingly, this study proposed to combine convolutional neural network (CNN)-based algorithms and traditional image morphological operations for crack detection and measurement. The proposed approach was tested on data from six images containing ten cracks of various sizes and shapes that were obtained from laboratory experiments in a controlled environment. The proposed methodology achieved an average F1 score of 0.93, with 88.17 percent accuracy in crack length measurements and 94.40 percent accuracy in width measurements. Future research will focus on fine-tuning the proposed crack detection and measurement methodology and will evaluate it with a set of images acquired from a full-scale structure such as a bridge or a building.