A grid‐based classification and box‐based detection fusion model for asphalt pavement crack

Existing crack recognition methods based on deep learning often face difficulties when detecting cracks in complex scenes such as brake marks, water marks, and shadows. The inadequate amount of available data can be primarily attributed to this factor. To address this issue, a controllable generative model of pavement cracks is proposed that can generate crack images in complex scenes by leveraging background images and crack mask images. The proposed model, the crack diffusion model (CDM), is based on the diffusion model network, which enables better control over the position and morphology of cracks by adjusting the conditional input of cracks. Experiments show that CDM has several advantages, including high definition, controllability, and sensitivity to narrow cracks. Utilizing CDM to create a synthetic crack data set in complex scenes resulted in substantial improvements of crack detection and segmentation. The method proposed in this study can effectively alleviate the effort required for data acquisition and labeling, especially in complex scenes.

Section: Introductionmentioning

confidence: 99%

A controllable generative model for generating pavement crack images in complex scenes

Zhang,

Qian,

Zhou

et al. 2024

“…Kim & Cho, 2019). Li et al (2023) proposed a road crack detection method combining localization and classification networks to efficiently estimate crack extension. Segmentation is important for obtaining detailed crack information such as bifurcations and widths.…”

Section: Introductionmentioning

confidence: 99%

Road crack detection interpreting background images by convolutional neural networks and a self‐organizing map

Yamaguchi,

Mizutani

2023

The presence of road cracks is an important indicator of damage. Deep learning is a prevailing method for detecting cracks in road surface images because of its detection ability. Previous research works focused on supervised convolutional neural networks (CNNs) without non‐crack features or unsupervised crack analysis with limited accuracies. The novelty of this study is the addition of background classification. By increasing the number of non‐crack categories, CNNs are driven to learn non‐crack features and improve crack detection performances. Non‐crack images are preprocessed, and their features are extracted in an unsupervised way by a deep convolutional autoencoder. A self‐organizing map clusters features to obtain non‐crack categories. This study focusses on classification though the method can be adopted in parallel with the latest segmentation algorithms. Using common road crack datasets, modified deep CNN models significantly improved accuracy by 1%–4% and f‐measure by 3%–8%, compared to previous models. The modified visual geometry group (VGG) 16 showed the top‐level performance, 96% accuracy and 84%–85% f‐measure. The models drastically reduced false detection cases while maintaining their crack detection abilities.

“…Chen et al., 2017), and other outstanding algorithms (Garcia‐Garcia et al., 2017). However, limited by some characteristics of crack images (i.e., small size, class imbalance, and random distribution of the shape), it is not effective to directly transfer or fine‐tune these algorithms that were originally developed for natural scene data segmentation to the scene of crack segmentation (Gao & Mosalam, 2018; Li et al., 2022). Therefore, most of the recent works were mainly focused on improving the accuracy of the algorithms in crack identification by combining some novel technologies inspired by the field of computer science (Hassanpour et al., 2019), including multi‐scale feature fusion (Hoskere et al., 2018), attention mechanisms (Qu et al., 2021), joint learning loss (Chu et al., 2022), and so forth.…”

Section: Introductionmentioning

confidence: 99%

“…There are two key factors that affect the performance of the aforementioned crack inspection methods, namely, the imaging quality of the captured crack image and the effectiveness of the crack detection algorithm (Guo et al., 2020; Li et al., 2022; Mirzaei & Adeli, 2018). For imaging quality, a much higher resolution can be achieved for captured images with the rapid development of display devices and photographic hardware.…”

Section: Introductionmentioning

confidence: 99%

Implicit function‐based continuous representation for meticulous segmentation of cracks from high‐resolution images

Chu

Long

Guo

et al. 2023

High‐resolution (HR) crack images offer more detailed information for evaluating the structural condition and formulating effective maintenance or rehabilitation plans. However, the meticulous segmentation of HR crack images has been a challenge due to the limitations of mainstream deep learning algorithms that extract features in a discrete manner, as well as the constraints of computing resources. To address this issue, a novel implicit function‐integrated architecture, called the crack continuous refinement network (CCRN), was proposed for meticulous segmentation of cracks from HR images using a continuous representation manner. First, a crack continuous alignment module with a position encoding function was proposed to encode the tiny crack pixels that are easily lost in the sampling process. Then, a lightweight decoder embedded with implicit functions was customized to recover crack details from the aligned latent features and continuous position encoding information. Afterward, the gap between low‐resolution training images and HR inference results was bridged by the proposed continuous inference strategy. Finally, the robustness and practicability of the well‐trained CCRN were demonstrated by a parallel comparison and an unmanned aerial vehicle‐based field experiment.