The Development of a Rebar-Counting Model for Reinforced Concrete Columns: Using an Unmanned Aerial Vehicle and Deep-Learning Approach

Inspecting the number of rebars in each column of a reinforced concrete (RC) structure is a significant task that must be undertaken during the rebar inspection process. Conventionally, counting the rebars has relied on a manual inspection carried out by visiting inspectors. However, this approach is very time-consuming, labor-intensive, and poses a potential safety risk. Previous studies have focused on the applications of counting the rebars for a production line and/or warehouse, using vision-based methods. Therefore, this study aims to propose an innovative approach incorporating the use of an unmanned aerial vehicle (UAV) on real construction sites to count the rebars automatically. For analyzing the images, robust object detection methods based on deep learning (Faster R-CNN, R-FCN, SSD 300, SSD500, YOLOv5, and YOLOv6) were developed. A total of 384 models generated from six different methods were trained and implemented using data sets based on the original and augmented images with adjustments made for the hyperparameters. In a test, the best optimized model based on Faster R-CNN produced an accuracy of 94.61% at AP50. In addition, video testing demonstrated a coverage of up to 32 frames per second in the experimental environment, suggesting that this method has potential for real-time application.

Drones provide an efficient way to monitor the number of rebars in reinforced columns by capturing still images or video footage. However, manually counting the rebars from this data in the form of images is both time-consuming and laborious. This research therefore develops an AI-driven technique, based on deep learning, designed to automate the process. In the experiment, the approach that was developed achieved an accuracy rate of 94.61% under diverse conditions on real construction sites, including nonuniform illumination and complex backgrounds (e.g., scaffolding and molding). Nevertheless, there is potential for further improvement in certain scenarios (e.g., where there are shadows in high-illumination images, or similar objects close to the rebars). In addition, video testing demonstrated that the system could process up to 32 frames per s. Despite its limitations, the method developed in this research could be put to practical use on construction sites, except in those scenarios where it showed a lower rate of accuracy. Moreover, as 30 frames per s is often regarded as equivalent to real-time, it would also be feasible to use it for video analytics’ applications such as real-time monitoring and progress tracking.