Please login first
Real-time concrete crack detection and instance segmentation using deep transfer learning
* 1 , 1 , 2 , 3
1  Central Queensland University, Australia
2  University of the sunshine coast
3  Federation University, Australia


Concrete based civil infrastructure such as bridges, tunnels and dams undergo structural deterioration due to weathering, corrosion, thermal cycles, and carbonation. Cracks on concrete surfaces are often identified as an early indication of possible future structural failures which could be catastrophic if unattended. Therefore, it is of utmost importance to inspect concrete structures frequently for cracks to initiate any proactive measures to avoid further damage. Visual inspection of larger civil structures using remotely controlled drones has become popular in recent years. The recorded video footages from these inspection rounds are manually watched to detect any cracks. This is a highly time-consuming process and largely depends on surveyor’s experience and the knowledge which adds an extra subjective bias to the final qualitative analysis.

In this paper, we demonstrate that deep transfer learning can be used to train an object detection model to automatically identify cracks with segmentation masks to localize cracks on images collected from video inspections. We specifically looked at YOLACT: a real-time instant segmentation algorithm which outperformed other existing algorithms in speed and accuracy in the COCO object detection dataset and used it to train deep learning model on a small dataset of concrete crack images. Instance segmentation allowed us to detect localized multiple cracks on the same image which may provide extra information to predict the propagation of cracks. Real-time detection is vital as this will enable active inspection by autonomously steering the drone along the cracks. Also, the drone can be navigated to closely look at the detected cracks. To train the crack detection model, we built a dataset by collecting images from a publicly available dataset and manually annotating segmentation mask for each crack. The transfer learning approach helped us to train the network on a smaller dataset with the high-level features extracted from the COCO dataset. The test on the trained model achieved a precision value of more than 90% and a recall value of more than 75%.

Keywords: Concrete cracks ; deep learning; visual inspection; infrastructure monitoring