Two-stage convolutional neural network for road crack detection and segmentation

Nguyen, Nhung Hong Thi; Perry, Stuart; Bone, Don; Le, Hung M.; Nguyễn, Thúy Thị

doi:10.1016/j.eswa.2021.115718

Cited by 54 publications

(22 citation statements)

References 21 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In detection phase, the existence of crack is determined by applying analytical or logical methods such as Otsu [28] methods, statistical approaches, and threshold methods. Many of road crack detection researches can be assigned into this category that actually determines the existence of cracks by incorporating various segmentation techniques to improve image quality or generates crack regions from the input image in the form of bounding boxes [35][36][37]39], but they does not actually determine the individual type of cracks. There are researches that utilize additional types of input data such as acoustic-sensor data [38] or 3D scanned data [40] to better detect cracks hidden below the surface.…”

Section: Crack Detection Using Deep Learningmentioning

confidence: 99%

Assessing severity of road cracks using deep learning-based segmentation and detection

Kim

2022

J Supercomput

View full text Add to dashboard Cite

For the successful operation of pavement management system, it is necessary to automate the detection, classification, and severity assessment of road cracks, which are bottlenecks in the entire process. Although good results for the detection and classification of road cracks have been published in many related studies, the number of crack types detected is still insufficient for actual field use, and studies on crack severity assessment are difficult to find. In this study, the number of crack types are expanded to five types (alligator crack, longitudinal crack, transverse crack, pothole, and patching) to meet the needs of fieldwork, and the assessment of crack severity is also included in the proposed model. In this system constructed using SqueezeNet, U-Net, and Mobilenet-SSD models together, an accuracy of 91.2% has been achieved for both crack type and severity assessment. The authors have performed segmentation of the input images using separately trained U-Nets for linear cracking and area cracking to improve object detection performance and automate crack severity assessment. With reference to the system presented in this study, it is expected that an automated pavement management system that better reflects each country’s requirements for various crack types and severity standards is possible.

show abstract

Section: Crack Detection Using Deep Learningmentioning

confidence: 99%

Assessing severity of road cracks using deep learning-based segmentation and detection

Kim

2022

J Supercomput

View full text Add to dashboard Cite

show abstract

“…Point cloud features are extracted from the same model that is used from Point-pixel combination. Image features are extracted from a crack detection model that was pre-trained in a previous model for crack detection [33]. This model is better suited for the method presented in this paper than other existing models for image features extraction because it was trained from a crack dataset.…”

Section: A Combining Images and Point Cloudsmentioning

confidence: 99%

Combination of Images and Point Clouds in a Generative Adversarial Network for Upsampling Crack Point Clouds

et al. 2022

Self Cite

View full text Add to dashboard Cite

Point cloud data of cracks can be used for various purposes such as crack detection, depth calculation and crack segmentation. Upsampling low-density point clouds can help to improve the performance of those tasks. Building on existing methods that upsample point clouds from low-resolution point cloud input, to improve feature definition, this paper proposes a new method for upsampling low-density point clouds using a combination of these point clouds and corresponding 2D images of the original objects as input data. We use an architecture based on Generative Adversarial Networks (GAN) for training input point clouds with additional information from the corresponding 2D images. The key idea is to exploit features from both 2D images and point clouds to enrich point clouds in both the training and testing phases. Our method takes advantage of the combination of 2D images and point clouds using a GAN framework. Experimental results show our proposed method achieves a higher effectiveness compared with previous upsampling methods.

show abstract

“…Road crack segmentation methods can be divided into two categories: traditional digital image processing and machine learning-based approaches. In the past five years, machine learning-based crack segmentation methods have repeatedly used advanced stateof-the-art techniques [9,13] and have replaced traditional methods. In this section, we review works involving convolutional neural networks for solving network architecture and imbalance problems.…”

Section: Related Workmentioning

confidence: 99%

“…A two-stage convolutional neural network [13] was proposed for road crack detection and segmentation in images at the pixel level. The first stage serves to remove noise or artifacts and isolate the potential cracks in a small area via a classification network that is composed of five-layer convolutional neural networks and two fully connected (FC) layers.…”

Section: Imbalance Problemmentioning

confidence: 99%

See 1 more Smart Citation

A Recurrent Adaptive Network: Balanced Learning for Road Crack Segmentation with High-Resolution Images

et al. 2022

View full text Add to dashboard Cite

Road crack segmentation based on high-resolution images is an important task in road service maintenance. The undamaged road surface area is much larger than the damaged area on a highway. This imbalanced situation yields poor road crack segmentation performance for convolutional neural networks. In this paper, we first evaluate the mainstream convolutional neural network structure in the road crack segmentation task. Second, inspired by the second law of thermodynamics, an improved method called a recurrent adaptive network for a pixelwise road crack segmentation task is proposed to solve the extreme imbalance between positive and negative samples. We achieved a flow between precision and recall, similar to the conduction of temperature repetition. During the training process, the recurrent adaptive network (1) dynamically evaluates the degree of imbalance, (2) determines the positive and negative sampling rates, and (3) adjusts the loss weights of positive and negative features. By following these steps, we established a channel between precision and recall and kept them balanced as they flow to each other. A dataset of high-resolution road crack images with annotations (named HRRC) was built from a real road inspection scene. The images in HRRC were collected on a mobile vehicle measurement platform by high-resolution industrial cameras and were carefully labeled at the pixel level. Therefore, this dataset has sufficient data complexity to objectively evaluate the real performance of convolutional neural networks in highway patrol scenes. Our main contribution is a new method of solving the data imbalance problem, and the method of guiding model training by analyzing precision and recall is experimentally demonstrated to be effective. The recurrent adaptive network achieves state-of-the-art performance on this dataset.

show abstract

Two-stage convolutional neural network for road crack detection and segmentation

Cited by 54 publications

References 21 publications

Assessing severity of road cracks using deep learning-based segmentation and detection

Assessing severity of road cracks using deep learning-based segmentation and detection

Combination of Images and Point Clouds in a Generative Adversarial Network for Upsampling Crack Point Clouds

A Recurrent Adaptive Network: Balanced Learning for Road Crack Segmentation with High-Resolution Images

Contact Info

Product

Resources

About