CrackU‐net: A novel deep convolutional neural network for pixelwise pavement crack detection

Ju Huyan,Wei Li,Susan Tighe,Zhengchao Xu,Junzhi Zhai

DOI: https://doi.org/10.1002/stc.2551

2020-03-18

Structural Control and Health Monitoring

Abstract:<p>Periodic road crack monitoring is an essential procedure for effective pavement management. Highly efficient and accurate crack measurements are key research topics in both academia and industry. Automatic methods gradually replaced traditional manual surveys for more reliable evaluation outputs and better efficiency, whereas the devices are not available to all functional classes of pavements and different departments considering the high cost versus the limited budget. Recently, the widespread use of smartphones and digital cameras made it possible to collect pavement surface crack images at an affordable price in easier ways. However, the qualities of these crack images are diversely influenced by the noises from pavement background, roadways, and so forth. Thus, traditional methods usually fail to extract accurate crack information from pavement images. Therefore, this research proposes a state‐of‐the‐art pixelwise crack detection architecture called CrackU‐net, which is featured by its utilization of advanced deep convolutional neural network technology. CrackU‐net achieved pixelwise crack detection through convolution, pooling, transpose convolution, and concatenation operations, forming the "U"‐shaped model architecture. The model is trained and validated by 3,000 pavement crack images, in which 2,400 for training and 600 for validating, using the Adam algorithm. CrackU‐net has the performance of loss = 0.025, accuracy = 0.9901, precision = 0.9856, recall = 0.9798, and F‐measure = 0.9842 with learning rate of <span>10⁻²</span>. Meanwhile, the false‐positive crack detection problem is avoided in CrackU‐net. Therefore, CrackU‐net outperforms both traditional approaches and fully convolutional network (FCN) and U‐net for pixelwise crack detections.</p>

construction & building technology,engineering, civil,instruments & instrumentation

What problem does this paper attempt to address?

The problem this paper attempts to address is: how to achieve efficient and accurate pixel-level crack detection in pavement crack detection. Traditional manual methods are time-consuming, labor-intensive, and not very accurate, while existing automated methods, although improving efficiency, still have false detection issues when dealing with images from different sources and qualities, especially in mistaking background noise for cracks. Therefore, the researchers propose a new architecture based on deep convolutional neural networks—CrackU-net, aiming to overcome the limitations of existing methods and improve the accuracy and robustness of crack detection.

Specifically, CrackU-net addresses these issues in the following ways:
1. **End-to-end deep learning architecture**: CrackU-net is an end-to-end deep learning model that can directly extract crack features from raw images and perform pixel-level crack detection.
2. **Reducing false detection issues**: By comprehensively utilizing multi-layer feature information, CrackU-net can effectively reduce false detection issues, particularly avoiding mistaking background noise for cracks.
3. **Flexibility and customizability**: The near-symmetric structure of CrackU-net provides high flexibility, allowing customization according to specific needs.

In summary, the main goal of this paper is to develop an efficient pixel-level pavement crack detection model to improve the accuracy and reliability of crack detection.