Search

The automation of bridge disease detection necessitates the time-consuming, labor-intensive manual detection process and the limitations of traditional image segmentation methods, such as inadequate denoising effects and insufficient continuity in crack segmentation. This paper proposes a rapid detection and information feedback approach based on an enhanced Convolutional Neural Network (CNN) model to tackle these issues in bridge crack width measurement and information processing. To improve efficiency and accuracy in bridge safety monitoring, the training data is constructed by the bridge image library and network crack through the refined preprocessing and image segmentation techniques applied to these images, key features of cracks are identified and extracted to enhance the capability for crack identification. For crack assessment, the maximum internal tangent circle method is employed to accurately measure the width of bridge abutment cracks. The effectiveness of our model was verified through both fixed-point detection and Unmanned Aerial Vehicle (UAV) dynamic detection, ensuring comprehensive and accurate data collection. This dual validation strategy shows that our model substantiates the wide applicability across various scenarios, and the non-contact crack measurement technique achieves a precision of 0.01 mm, demonstrating the effectiveness and accuracy of this streamlined CNN model in accurately assessing crack width.

Abstract With the rapid advancement of highway construction, the maintenance of highway infrastructure has become particularly vital. During highway maintenance, the effective detection of tiny road surface cracks helps to extend the lifespan of roads and enhance traffic efficiency and safety. To elevate the performance of existing road detection models, the CrackTinyNet (CrTNet) algorithm is specifically proposed for detecting tiny road surface cracks. This algorithm utilizes the novel BiFormer general visual transformer, designed expressly for tiny objects, and optimizes the loss function to a normalized Wasserstein distance loss function. It replaces traditional downsampling with Space‐to‐Depth Conv to prevent the excessive loss of tiny object information in the network structure. To highlight the model's advantage in detecting tiny road cracks, ablation experiments and comparison trials were conducted with mainstream deep learning models for crack detection. The results of the ablation experiments show that, compared to the baseline, CrTNet improved the Mean Average Precision (MAP) by 0.22. When compared to other network models suitable for road detection, these results exhibited an improvement of over 8.9%. In conclusion, the CrTNet proposed in this study enables a more accurate detection of tiny road cracks, playing a significant role in the advancement of intelligent traffic management.

The importance of the parameters of any optimization algorithm, especially meta-heuristic algorithms that have been created to simplify the solution of optimization problems, is inevitable. The optimal values of these parameters, which generally depend on the specifics of the problem in question, have a significant impact on the performance of the mentioned algorithms and a better search of the solution space. Parameters selection of them will play an important role in performance and efficiency of the algorithms. This article examines the capability of various optimization algorithms and suggests dual hybrid optimization algorithms are named PSO-FA, PSO-GA, PSO-GWO, for solving the problem of computing the depth and location of cracks in cantilever beams. The performance of Particle swarm optimization (PSO), Genetic algorithm (GA), Grey wolf optimization (GWO), Firefly algorithm (FA), and hybrid of them base on PSO optimizer to determine the location and depth of crack for cantilever beam are proposed. These suggested algorithms are optimization algorithms based on intelligent optimization. So, the performance of these algorithms are analyzed when the control parameters vary.

Purpose – The purpose of this study is to study the quantitative evaluation method of contact wire cracks by analyzing the changing law of eddy current signal characteristics under different cracks of contact wire of high-speed railway so as to provide a new way of thinking and method for the detection of contact wire injuries of high-speed railway. Design/methodology/approach – Based on the principle of eddy current detection and the specification parameters of high-speed railway contact wires in China, a finite element model for eddy current testing of contact wires was established to explore the variation patterns of crack signal characteristics in numerical simulation. A crack detection system based on eddy current detection was built, and eddy current detection voltage data was obtained for cracks of different depths and widths. By analyzing the variation law of eddy current signals, characteristic parameters were obtained and a quantitative evaluation model for crack width and depth was established based on the back propagation (BP) neural network. Findings – Numerical simulation and experimental detection of eddy current signal change rule is basically consistent, based on the law of the selected characteristics of the parameters in the BP neural network crack quantitative evaluation model also has a certain degree of effectiveness and reliability. BP neural network training results show that the classification accuracy for different widths and depths of the classification is 100 and 85.71%, respectively, and can be effectively realized on the high-speed railway contact line cracks of the quantitative evaluation classification. Originality/value – This study establishes a new type of high-speed railway contact wire crack detection and identification method, which provides a new technical means for high-speed railway contact wire injury detection. The study of eddy current characteristic law and quantitative evaluation model for different cracks in contact line has important academic value and practical significance, and it has certain guiding significance for the detection technology of contact line in high-speed railway.

Abstract The accurate detection of tunnel lining cracks and prompt identification of their primary causes are critical for maintaining tunnel availability. The advancement of deep learning, particularly in the domain of convolutional neural network (CNN) for image segmentation, has made tunnel lining crack detection more feasible. However, the CNN-based technique for tunnel lining crack detection commonly prioritizes increasing algorithmic complexity to enhance detection accuracy, posing a challenge in balancing the accuracy of detection and the efficiency of the algorithm. Motivated by the superior performance of Unet in image segmentation, this paper proposes a lightweight tunnel lining crack detection model named Mini-Unet, which refined the Unet architecture and utilized depthwise separable convolutions (DSConv) to replace some standard convolution layers. In the optimization of the proposed model parameters, applying a hybrid loss function that integrated dice loss and cross-entropy loss effectively tackled the imbalance between crack and background categories. Several models were set up to contrast with Mini-Unet and the experimental results were analyzed. Mini-Unet achieves a mean intersection over union (MIoU) of 60.76%, a mean precision of 84.18%, and a frame per second (FPS) of 5.635, respectively. Mini-Unet outperforms several mainstream models, enabling rapid detection while maintaining identified accuracy and facilitating the practical application of AI power for real-time tunnel lining crack detection.

Abstract Unsupervised anomaly detection, often approached as a one‐class classification problem, is a critical task in computer vision. Knowledge distillation has emerged as a promising technique for enhancing anomaly detection accuracy, especially with the advent of reverse distillation networks that employ encoder–decoder architectures. This study introduces a novel reverse knowledge distillation framework known as RDMS, which incorporates a pretrained teacher encoding module, a multi‐level feature fusion connection module, and a student decoding module consisting of three independent decoders. RDMS is designed to distill distinct features from the teacher encoder, mitigating overfitting issues associated with similar or identical teacher–student structures. The model achieves an average of 99.3% image‐level AUROC and 98.34% pixel‐level AUROC on the MVTec‐AD dataset and demonstrates state‐of‐the‐art performance on the more challenging BTAD dataset. The RDMS model's high accuracy in anomaly detection and localization underscores the potential of multi‐student reverse distillation to advance unsupervised anomaly detection capabilities. The source code is available at https://github.com/zihengchen777/RDMS

Underwater concrete structure crack detection and structural health condition assessment based on image processing is a challenging task. The complex underwater environment and severe image degradation seriously affect the accuracy of crack detection. To solve these problems, a monocular vision and image-enhanced fractal-based fractal science based on computer vision and image processing techniques are proposed to carry out a non-contact detection study of underwater concrete cracks. In this study, a four-level structural health condition was established to assist in underwater crack measurement and safety assessment. The box-counting method was used as a practical tool to calculate the fractal dimension. To verify the effective distance of the algorithm, three distances of 0.5 m,0.8 m, and 1.2 m were set. The results show that the method proposed in this study can effectively detect cracks in submerged concrete members within 0.6 m and help managers correctly determine the health of the structure using the fractal dimension.

Abstract Accurate crack detection is crucial for maintaining pavement integrity, yet manual inspections remain labor-intensive and prone to errors, underscoring the need for automated solutions. This study proposes a novel crack segmentation approach utilizing advanced visual models, specifically Detectron2 and the Segment Anything Model (SAM), applied to the CFD and Crack500 datasets, which exhibit intricate and diverse crack patterns. Detectron2 was tested with four configurations—mask_rcnn_R_50_FPN_3x, mask_rcnn_R_101_FPN_3x, faster_rcnn_R_50_FPN_3x, and faster_rcnn_R_101_FPN_3x—while SAM was compared using Focal Loss, DiceCELoss, and DiceFocalLoss. SAM with DiceFocalLoss outperformed Detectron2, achieving mean IoU scores of 0.69 and 0.59 on the CFD and Crack500 datasets, respectively. The integration of Detectron2 with faster_rcnn_R_101_FPN_3x and SAM using DiceFocalLoss involves generating bounding boxes with Detectron2, which serve as prompts for SAM to produce segmentation masks. This approach achieves mIoU scores of 0.83 for CFD dataset and 0.75 for Crack500 dataset. These results highlight the potential of combining foundation models with Detectron2 for advancing crack detection technologies, offering valuable insights for enhancing highway maintenance systems.

Concrete structure damage and severity are determined by examining the width of cracks in various forms. This categorisation of cracks is based on their specific types, which is crucial for engineers and professionals. It allows them to efficiently prioritise maintenance and repair efforts, extending the lifespan of concrete structures. Thus, the classification of cracks is based on the type of cracks needed. This study combines a wall-climbing robot equipped with imaging capabilities to automate the detection and classification of cracks in concrete surfaces. We used machine learning to classify the crack orientation and OTSU to segment the crack shapes. Based on this study, four experiments were carried out using machine learning methods to classify types of cracks, which are SVM, Random Forest, KNN, and Decision Tree. These experiments were categorised using multiclass classification with types of the orientation of crack such as Not crack, Crocodile, Transverse, and Longitudinal. The classification one-class results show that the Decision Tree achieved 86.50%, SVM 99.50%, Random Forest 97%, and KNN 40%. In multiclass classification, Decision Tree achieved 64%, Random Forest 80%, and KNN 37%. The higher accuracy from SVM is achieved at 87%.

Abstract In the inspection and diagnosis of concrete construction, crack detection is highly recommended in the earliest phases to prevent any potential risks later. However, the flaws in concrete surfaces cannot be reliably and effectively identified using traditional crack detection techniques. The suggested algorithm is a supportive tool for agents or authorities to use in crack detection mechanisms to monitor and assess the current condition of buildings or bridges. The researchers aim to establish an intelligent model for automatic crack detection on different concrete surfaces based on image processing technology. Three different concrete surfaces—bridge decks, walls, and concrete cubes—are used to test the model. A subset of the public dataset of bridge decks and walls from SDNET (2018) and 150*150*150 mm of concrete cubes taken from the material laboratory of the faculty of engineering at Ain Shams University are applied to the model. The model F1-score measures are 98.87%, 97.43%, and 74.11% for detecting cracks in bridges, walls, and concrete cubes, respectively. The validation of the applicability of the suggested novel approach is based on a comparison with recent methods for crack recognition. The contribution of this study is that it could be applied to detect cracks efficiently on three different types of concrete surfaces, including uneven concrete surfaces, random noise, voids, dents, colour changes, and stain marks. The proposed method is transparent in its workflow and has a lower computational cost compared with deep learning frameworks. Thus, the outcomes of this model demonstrate its effectiveness in concrete defect field investigation.

This study explores a non-destructive testing (NDT) method for crack detection using a two-stage convolutional neural network (CNN) model, incorporating a combination of AlexNet and YOLO models through transfer learning. Crack detection is pivotal for assessing structural integrity and ensuring timely maintenance interventions. The developed model was rigorously tested in simulated environments and through physical experimentations with the use of a UAV to evaluate its effectiveness. A 2-stage model, based on AlexNet and YOLO, was developed for crack classification and segmentation. The developed model leveraged transfer learning to address limitations from traditional CNN models. A known dataset was used to evaluate the developed model, benchmarking it against other models. The classification network achieved an accuracy rate exceeding 90%, while the segmentation network successfully identified and delineated cracks in 85.71% of the images. Finally, the developed model was deployed using a UAV to perform crack detection and segmentation in a controlled environment. These results underscore the model's proficiency in both detecting and segmenting structural cracks, highlighting its potential as a reliable tool for enhancing the maintenance and safety of architectural structures. Doi: 10.28991/HIJ-2024-05-03-010 Full Text: PDF

Abstract U-net, a fully convolutional network-based image segmentation method, has demonstrated widespread adaptability in the crack segmentation task. The combination of the semantically dissimilar features of the encoder (shallow layers) and the decoder (deep layers) in the skip connections leads to blurry features map and leads to undesirable over- or under-segmentation of target regions. Additionally, the shallow architecture of the U-Net model prevents the extraction of more discriminatory information from input images. This paper proposes a Residual Sharp U-Net (RS-Net) architecture for crack segmentation and severity assessment in pavement surfaces to address these limitations. The proposed architecture uses residual block in the U-Net model to extract a more insightful representation of features. In addition to that, a sharpening kernel filter is used instead of plain skip connections to generate a fine-tuned encoder features map before combining it with decoder features maps to reduce the dissimilarity between them and smoothes artifacts in the network layers during early training. The proposed architecture is also integrated with various morphological operations to assess the severity of cracks and categorize them into hairline, medium, and severe labels. Experiments results demonstrated that the RS-Net model has promising segmentation performance, outperforming earlier U-Net variations on testing data for crack segmentation and severity assessment, with a promising accuracy (>0.97)

Abstract Cracks in tunnel lining structures constitute a common and serious problem that jeopardizes the safety of traffic and the durability of the tunnel. The similarity between lining seams and cracks in terms of strength and morphological characteristics renders the detection of cracks in tunnel lining structures challenging. To address this issue, a new deep learning-based method for crack detection in tunnel lining structures is proposed. First, an improved attention mechanism is introduced for the morphological features of lining seams, which not only aggregates global spatial information but also features along two dimensions, height and width, to mine more long-distance feature information. Furthermore, a mixed strip convolution module leveraging four different directions of strip convolution is proposed. This module captures remote contextual information from various angles to avoid interference from background pixels. To evaluate the proposed approach, the two modules are integrated into a U-shaped network, and experiments are conducted on Tunnel200, a tunnel lining crack dataset, as well as the publicly available crack datasets Crack500 and DeepCrack. The results show that the approach outperforms existing methods and achieves superior performance on these datasets.

Buildings with cracks are extremely hazardous because they have the potential to cause destruction. Numerous occupants of structures such as houses and buildings are at risk when cracks appear. There are numerous techniques for identifying fractures in structures, including visual inspection, tool use, and expert inspection. The present study employed computer vision, a form of artificial intelligence, to detect cracks in buildings. The main objective of this research is to construct a prototype capable of real-time monitoring of cracks in building walls. This research makes use of a methodology that combines machine learning and deep learning. Machine learning is employed in the classification process, whereas deep learning is utilized for the extraction of features. This research employs MobileNetV2 as its deep learning architecture and K-NN, Naive Bayes, SVM, XGBoost, and Random Forest as its machine learning classifiers. Test results show that when dividing the 80:20 dataset, XGBoost algorithms can produce the highest accuracy, sensitivity, and specificity values of 99%. Tests in the real environment are performed by deploying Raspberry Pi. Test results show that the prototype can detect cracks inthe structure surfaceat a distance of 10 meters in a bright environment. The crack detection process is carried out in real time at an average speed of 42fps.

During the industrial pecan shelling process, kernels are often damaged. To address this problem, a study is conducted to experimentally determine improved impactor geometries for end-to-end pecan cracking. Four impactors of varying internal angles (from 30° to 52.5°, in increments of 7.5°) are tested. After cracking, the pecans are passed through an image analysis software designed to detect and measure cracks in their shells. These measurements help classify each pecan into one of four categories: under crack, standard crack, ideal crack, or over crack. Cracked and ideally cracked pecans are preferred for their processability, so the impactor geometries are then evaluated based on their ability to maximize these crack types across the widest impact energy range. For the four impactors tested, the 30° impactor is found to more consistently produce preferred cracks in a larger energy range relative to the other impactors.

Abstract Technology offers a lot of potential that is being used to improve the integrity and efficiency of infrastructures. Crack is one of the major concerns that can affect the integrity or usability of any structure. Oftentimes, the use of manual inspection methods leads to delays which can worsen the situation. Automated crack detection has become very necessary for efficient management and inspection of critical infrastructures. Previous research in crack detection employed classification and localization-based models using Deep Convolutional Neural Networks (DCNNs). This study suggests and compares the effectiveness of transfer learned DCNNs for crack detection as a classification model and as a feature extractor to overcome this restriction. The main objective of this paper is to present various methods of crack detection on surfaces and compare their performance over 3 different datasets. Experiments conducted in this work are threefold: initially, the effectiveness of 12 transfer learned DCNN models for crack detection is analyzed on three publicly available datasets: SDNET, CCIC and BSD. With an accuracy of 53.40%, ResNet101 outperformed other models on the SDNET dataset. EfficientNetB0 was the most accurate (98.8%) model on the BSD dataset, and ResNet50 performed better with an accuracy of 99.8% on the CCIC dataset. Secondly, two image enhancement methods are employed to enhance the images and are transferred learned on the 12 DCNNs in pursuance of improving the performance of the SDNET dataset. The results from the experiments show that the enhanced images improved the accuracy of transfer-learned crack detection models significantly. Furthermore, deep features extracted from the last fully connected layer of the DCNNs are used to train the Support Vector Machine (SVM). The integration of deep features with SVM enhanced the detection accuracy across all the DCNN-dataset combinations, according to analysis in terms of accuracy, precision, recall, and F1-score.

Abstract The railway rivet is one of the most important and easily damaged parts of the connection. If rivets develop cracks during the production process, their load-bearing capacity will be reduced, thereby increasing the risk of failure. Fluorescent magnetic particle flaw detection (FMPFD) is a widely used inspection method for train fasteners. Manual inspection is not only time-consuming but also prone to miss detection, therefore intelligent detection system has important application value. However, the fluorescent crack images obtained by FMPFD present challenges for intelligent detection, such as the dense, multi-scaled and uninstantiated cracks. In addition, there is limited research on fluorescent rivet crack detection. This paper adopts instance segmentation to achieve automatic cracks detection of rivets. A decentralized target center and low overlap rate labeling method is proposed, and a Gaussian-weighted correction post-processing method is introduced to improve the recall rate in the areas of dense cracks. An efficient channel spatial attention mechanism for feature extraction is proposed in order to enhance the detection of multi-scale cracks. For uninstantiated cracks, an improvement of crack detection in uninstantiated regions based on multi task feature learning is proposed, thoroughly utilizing the semantic and spatial features of the fluorescent cracks. The experimental results show that the improved methods are better than the baseline and some cutting-edge algorithms, achieving a recall rate and mAP0.5 of 86.4% and 90.3%. In addition, a single coil non-contact train rivet composite magnetization device is built for rivets that can magnetize different shapes of rivets and has universality.

The objective of this study is to develop and apply deep neural networks for the automation of crack detection in buildings. The methodology involved training the YOLO v8 network with images collected from the internet, aiming to identify and locate cracks in real time. The model obtained 80% accuracy in validation with images not used in training, despite performance limitations in Google Collab. These limitations included restrictions on the execution environment, and the model is specific to cracks. The originality of the tool lies in its relevance for the automated detection of cracks, with the potential to extend to other pathological manifestations. It is concluded that the application of deep neural networks offers an efficient solution for the identification of problems in buildings.

Abstract Crack formation is a common phenomenon in engineering structures, which can cause serious damage to the safety and health of these structures. An important method of ensuring the safety and health of engineered structures is the prompt detection of cracks. Image threshold segmentation based on machine vision is a crucial technology for crack detection. Threshold segmentation can separate the crack area from the background, providing convenience for more accurate measurement and evaluation of the crack condition and location. The segmentation of cracks in complex scenes is a challenging task, and this goal can be achieved by means of multilevel thresholding. The arithmetic-geometric divergence combines the advantages of the arithmetic mean and the geometric mean in probability measures, enabling a more precise capture of the local features of an image in image processing. In this paper, a multilevel thresholding method for crack image segmentation based on the minimum arithmetic-geometric divergence is proposed. To address the issue of time complexity in multilevel thresholding, an enhanced particle swarm optimization algorithm with local stochastic perturbation is proposed. In crack detection, the thresholding criterion function based on the minimum arithmetic-geometric divergence can adaptively determine the thresholds according to the distribution characteristics of pixel values in the image. The proposed enhanced particle swarm optimization algorithm can increase the diversity of candidate solutions and enhance the global convergence performance of the algorithm. The proposed method for crack image segmentation is compared with seven state-of-the-art multilevel thresholding methods based on several metrics, including RMSE, PSNR, SSIM, FSIM, and computation time. The experimental results show that the proposed method outperforms several competing methods in terms of these metrics.

Survey of road cracks in a timely, complete, and accurate way is pivotal to pavement maintenance planning. Motivated by the increasingly heavy task of identifying cracks, researchers have developed extensive crack segmentation models based on Deep learning (DL) methods with significantly different levels of accuracy, efficiency, and generalizing capacity. Although many of the models provide satisfying detection performance, why these models work still needs to be determined. The objective of this study is to survey recent advances in automated DL crack recognition and provide evidence for their underlying working mechanism. We first reviewed 54 DL crack recognition methods to summarize critical factors in these models. Then, we conducted a performance evaluation of fourteen famous semantic segmentation models using the quantitative metrics: F-1 score and mIoU. Then, the effective receptive field and class activation map of the included models are visualized to demonstrate the training results as qualitative evaluation. Based on the literature review and comparison results, larger kernel size, feature fusion, and attention module all contribute to the improvement of model performance. Striking a balance between increasing the effective receptive field and computational/memory efficiency is the key to designing DL crack segmentation models. Finally, some potential directions and suggestions for future development are provided, such as developing semi-supervised or unsupervised learning for the high cost of pixel-level labeling.

Abstract In concrete structures, surface cracks are an important indicator for assessing the durability and serviceability of the structure. Existing convolutional neural networks for concrete crack identification are inefficient and computationally costly. Therefore, a new Cross Swin transformer-skip (CSW-S) is proposed to classify concrete cracks. The method is optimized by adding residual links to the existing Cross Swin transformer network and then trained and tested using a dataset with 17,000 images. The experimental results show that the improved CSW-S network has an extended range of extracted image features, which improves the accuracy of crack recognition. A detection accuracy of 96.92% is obtained using the trained CSW-S without pretraining. The improved transformer model has higher recognition efficiency and accuracy than the traditional transformer model and the classical CNN model.

Complex structures can develop cracks and defects over time, which can compromise their long-term performance and safety. Structural Health Monitoring (SHM) systems are essential for detecting and measuring these defects by monitoring the load and deformation of the solid materials. This paper presents a simulation study of the frequency and strength of solid cylindrical bars made of aluminum and steel under different loads and crack conditions. Finite Element Method (FEM) and COMSOL Multiphysics software are used to perform the simulation, and a resonance model is used to analyze the results. The study investigates how cracks affect the frequency and deformation of the bars, and how different materials respond to load and bending. The results show that frequency varies linearly with load, cracks decrease the stiffness and increase the frequency at the crack location, and aluminum bars deform more than steel bars. The paper concludes that steel bars are more resistant to load and bending than aluminum bars for both cracked and uncracked case. Finally, it is found that steel bars are more resistant to load and bending than aluminum bars for both cracked and uncracked case.

Automatic crack detection in construction facilities is a challenging yet crucial task. However, existing deep learning (DL)-based semantic segmentation methods for this field are based on fully supervised learning models and pixel-level manual annotation, which are time-consuming and labor-intensive. To solve this problem, this paper proposes a novel crack semantic segmentation network using weakly supervised approach and mixed-label training strategy. Firstly, an image patch-level classifier of crack is trained to generate a coarse localization map for automatic pseudo-labeling of cracks combined with a thresholding-based method. Then, we integrated the pseudo-annotated with manual-annotated samples with a ratio of 4:1 to train the crack segmentation network with a mixed-label training strategy, in which the manual labels were assigned with a higher weight value. The experimental data on two public datasets demonstrate that our proposed method achieves a comparable accuracy with the fully supervised methods, reducing over 65% of the manual annotation workload.

Abstract Crack detection is a long-standing topic in structural health monitoring. Conventional damage detection techniques rely on intensive, time-consuming, resource-intensive intervention. The current trend of crack detection emphasizes using deep neural networks to build an automated pipeline from measured signals to damaged areas. This work focuses on the seismic-wave-based technique of crack detection for plate structures. Previous work proposed an encoder–decoder network to extract crack-related wave patterns from measured wave signals and predict crack existence on the plate. We extend previous work with extensive experiments on different network components and a data preprocessing strategy. The proposed methods are tested on an expanded crack detection dataset. We found that a robust backbone network, such as Densely Connected Convolutional Network (DenseNet) can effectively extract the features characterizing cracks of wave signals, and by using the reference wave field for normalization, the accuracy of detecting small cracks can be further improved.

As the expected lifespans of structures and road approaches, as well as the importance of road maintenance, increase globally, safety inspections have emerged as a crucial task. Nonetheless, the existing crack detection models focus on multi-scale feature loss and performance degradation in learning various types of cracks. We propose the Multi-Scale Parallel Attention U-Net (MSP U-Net) as a network designed for low-resolution images that considers the irregular characteristics of cracks. MSP U-Net applies a large receptive field flock to an attention U-Net, minimizing feature loss across multiple scales. Using the Crack500 dataset, our network achieved a mean intersection of union (mIoU) of 0.7752, outperforming the existing methods on low-resolution images.

Owing to the long-standing statement that natural frequency-based crack detection is not sensitive enough to localised damage to identify small cracks, many natural frequency-based crack detection methods are validated by detecting cracks of moderate size. However, a direct comparison between the crack severity causing a measurable natural frequency change and the crack severity reaching the initiation of crack propagation or leading to brittle fracture is constantly ignored. Without this understanding, it is debatable whether the presented crack detection methods are feasible in practical situations. Through natural frequency calculation and linear elastic fracture mechanics, this study is dedicated to filling the above gap in knowledge. To directly utilize the solution of stress intensity factor, common fracture toughness test specimens featuring a single-edge crack are used. These specimens are essentially cracked rectangular plates under uniform uniaxial tension. Considering the stress resultants obtained via the extended finite element method, the natural frequency of the loaded cracked plates is calculated using the Rayleigh–Ritz method incorporating corner functions. In addition, assuming the specimens as structural components under remote uniform tension, the development of critical load versus crack length is derived based on the solution of the stress intensity factor. Thus, critical crack lengths corresponding to a series of safety factors are obtained by equating service load with critical load. After obtaining natural frequencies of the cracked plates with critical crack lengths, the natural frequency drop caused by a critical crack can be computed. Hence, the critical crack length can be compared with the crack length when the frequency drop is measurable. It is found that the brittleness of the employed metals plays a vital role in the feasibility of natural frequency-based crack detection.

The You Only Look Once (YOLO) network is considered highly suitable for real-time object detection tasks due to its characteristics, such as high speed, single-shot detection, global context awareness, scalability, and adaptability to real-world conditions. This work introduces a comprehensive analysis of various YOLO models for detecting cracks in concrete structures, aiming to assist in the selection of an optimal model for future detection and segmentation tasks. The YOLO models are initially trained on a dataset containing both images with and without cracks, producing a generalized model capable of extracting abstract features beneficial for crack detection. Subsequently, transfer learning is employed using a dataset that reflects real-world conditions, such as occlusions, varying crack sizes, and rotations, to further refine the model. Crack detection in concrete remains challenging due to the wide variation in crack sizes, aspect ratios, and complex backgrounds. To achieve optimal performance, we test different versions of YOLO, a state-of-the-art single-shot detector, and aim to balance inference speed and mean average precision (mAP). Our results indicate that YOLOv10 demonstrates superior performance, achieving a mean average precision (mAP) of 74.52% with an inference time of 19.5 milliseconds per image, making it the most effective among the models tested.

Early identification of concrete cracks and multi-class detection can help to avoid future deformation or collapse in concrete structures. Available traditional detection and methodologies require enormous effort and time. To overcome such difficulties, current vision-based deep learning models can effectively detect and classify various concrete cracks. This study introduces a novel multi-stage deep learning framework for crack detection and type classification. First, the recently developed YOLOV10 model is trained to detect possible defective regions in concrete images. After that, a modified vision transformer (ViT) model is trained to classify concrete images into three main types: normal, simple cracks, and multi-branched cracks. The evaluation process includes feeding concrete test images into the trained YOLOV10 model, identifying the possible defect regions, and finally delivering the detected regions into the trained ViT model, which decides the appropriate crack type of those detected regions. Experiments are conducted using the individual ViT model and the proposed multi-stage framework. To improve the generation ability, multi-source datasets of concrete structures are used. For the classification part, a concrete crack dataset consisting of 12,000 images of three classes is utilized, while for the detection part, a dataset composed of various materials from historical buildings containing 1116 concrete images with their corresponding bounding boxes, is utilized. Results prove that the proposed multi-stage model accurately classifies crack types with 90.67% precision, 90.03% recall, and 90.34% F1-score. The results also show that the proposed model outperforms the individual classification model by 10.9%, 19.99%, and 19.2% for precision, recall, and F1-score, respectively. The proposed multi-stage YOLOV10-ViT model can be integrated into the construction systems which are based on crack materials to obtain early warning of possible future deformation in concrete structures.

Automatic crack detection is challenging, owing to the complex and thin topologies, diversity, and background noises of cracks. Inspired by the wavelet theory, we present an instance normalization wavelet (INW) layer and embed the layer into the deep model for segmentation. The proposed layer employs prior knowledge in the wavelets to capture the crack features and filter the high-frequency noises simultaneously, accelerating the convergence of model training. Furthermore, instance normalization in our layer is utilized to mitigate the feature differences, boosting the generalization capability. In addition, a fusion layer is added to merge the information across the different layers. The comparison experiments and ablation studies demonstrate that the INW layer steadily enhances recognition and convergence performance on the DeepCrack dataset and CRACK500 dataset.

The use of Artificial Intelligence (AI) to detect defects such as concrete cracks in civil and transport infrastructure has the potential to make inspections less expensive, quicker, safer and more objective by reducing the need for on-site human labour. One deployment scenario involves using a drone to carry an embedded device and camera, with the device making localised predictions at the edge about the existence of defects using a trained convolutional neural network (CNN) for image classification. In this paper, we trained six CNNs, namely Resnet18, Resnet50, GoogLeNet, MobileNetV2, MobileNetV3-Small and MobileNetV3-Large, using transfer learning technology to classify images of concrete structures as containing a crack or not. To enhance the model’s robustness, the original dataset, comprising 3000 images of concrete structures, was augmented using salt and pepper noise, as well as motion blur, separately. The results show that Resnet50 generally provides the highest validation accuracy (96% with the original dataset and a batch size of 16) and the highest validation F1-score (95% with the original dataset and a batch size of 16). The trained model was then deployed on an Nvidia Jetson Nano device for real-time inference, demonstrating its capability to accurately detect cracks in both laboratory and field settings. This study highlights the potential of using transfer learning on Edge AI devices for Structural Health Monitoring, providing a cost-effective and efficient solution for automated crack detection in concrete structures.

Cracks represent one of the most common types of damage in building structures and it is crucial to detect cracks in a timely manner to maintain the safety of the buildings. In general, tiny cracks require focusing on local detail information while complex long cracks and cracks similar to the background require more global features for detection. Therefore, it is necessary for crack detection to effectively integrate local and global information. Focusing on this, a local–global feature adaptive fusion network (LGFAF-Net) is proposed. Specifically, we introduce the VMamba encoder as the global feature extraction branch to capture global long-range dependencies. To enhance the ability of the network to acquire detailed information, the residual network is added as another local feature extraction branch, forming a dual-encoding network to enhance the performance of crack detection. In addition, a multi-feature adaptive fusion (MFAF) module is proposed to integrate local and global features from different branches and facilitate representative feature learning. Furthermore, we propose a building exterior wall crack dataset (BEWC) captured by unmanned aerial vehicles (UAVs) to evaluate the performance of the proposed method used to identify wall cracks. Other widely used public crack datasets are also utilized to verify the generalization of the method. Extensive experiments performed on three crack datasets demonstrate the effectiveness and superiority of the proposed method.

As a critical component of the transportation system, the safety of bridges is directly related to public safety and the smooth flow of traffic. This study addresses the aforementioned issues by focusing on the identification of bridge structure deterioration and the updating of finite element models, proposing a systematic research framework. First, this study presents a preprocessing method for bridge point cloud data and determines the parameter ranges for key algorithms through parameter tuning. Subsequently, based on the massive point cloud data, this research explores and optimizes the methods for identifying bridge cracks and spatial deformations, significantly enhancing the accuracy and efficiency of identification. On this basis, the particle swarm optimization algorithm is employed to optimize the key parameters in crack detection, ensuring the reliability and precision of the algorithm. Additionally, the study summarizes the methods for detecting bridge structural deformations based on point cloud data and establishes a framework for updating the bridge model. Finally, by integrating the results of bridge crack and deformation detection and combining Bayesian model correction and adaptive nested sampling methods, this research sets up the process for updating finite element model parameters and applies it to the analysis of actual bridge point cloud data.

In order to maintain serviceability and reliability of concrete structures, it is essential to assess their condition as concrete structures deteriorate in time. Cracks develop in concrete due to several reasons such as severe loading, environmental effects, chemical effects etc. and cause durability loss in the structure which may also lead to loss of stability. In this research, crack detection is realized by machine learning and an infrared image. The effects of infrared images on crack detection are confirmed by random forest algorithm to select useful explanatory features. Selected features are applied to random forest algorithm and neural network algorithm. Effective filters are selected as a feature selection technique to improve the accuracy. Crack detection is also conducted by U-Net with RGB and infrared images, and the detection characteristics are compared to conventional methods. The performance of two conventional machine learning methods, random forest and neural network, are evaluated based on F1 score and false positives. Applying selected features improves the accuracy of the crack detection from an infrared image. False positives decreased due to monitoring conditions and camera specifications in the infrared image. The most effective image processing filter is the blur filter for each algorithm. Comparing algorithms for crack detection using selected features, different accuracy values are obtained. U-Net enables more accurate crack detection compared to conventional methods. The number of false positives is reduced compare to conventional method. In the detection results by three algorithms, infrared image affects the balance of false negatives and false positives.

This study aims to recognize fluorescent excitation images of microcracks in concrete components through the adoption of an Attention Mechanism-based Deep Recurrent Neural Network (RNN) model, thereby enhancing the accuracy and efficiency of crack detection. Considering the significance of concrete crack detection and the limitations in efficiency and accuracy of existing methods, this paper proposes an innovative image processing technique that combines fluorescent excitation methods with deep learning models to achieve earlier and more accurate identification of concrete microcracks. C30 concrete specimens were prepared experimentally and treated with fluorescent solution spray. Fluorescent images of cracks under UV light excitation were collected and processed using a segmented attention mechanism deep RNN model. Various performance evaluation metrics, including mean Intersection over Union (mIoU), mean Image Intersection over Union (miIoU), mean Image Dice Coefficient (miDice), and F1 score, were employed to comprehensively assess the model's performance. The results demonstrate that the proposed model achieved significant effectiveness in concrete crack image recognition, showing higher mIoU, miIoU, miDice, and F1 scores compared to other representative deep learning models, thus proving its advantages in recognition accuracy and efficiency. Particularly, by introducing the segmented attention mechanism, the model could capture microcrack features more effectively, significantly improving the accuracy of crack identification. This method not only provides a new technical approach for the early detection of concrete cracks but also lays the foundation for further development of efficient and accurate crack detection technologies to accommodate more complex engineering application scenarios.

To address the issue of detecting complex-shaped cracks that rely on manual, which may result in high costs and low efficiency, this paper proposed a lightweight ground crack rapid detection method based on semantic enhancement. Firstly, the introduction of the Context Guided Block module enhanced the YOLOv8 backbone network, improving its feature extraction capability. Next, the incorporation of GSConv and VoV-GSCSP was introduced to construct a lightweight yet efficient neck network, facilitating the effective fusion of information from multiple feature maps. Finally, the detection head achieved more precise target localization by optimizing the probability around the labels. The proposed method was validated through experiments on the public dataset RDD-2022. The experimental results demonstrate that our method effectively detects cracks. Compared to YOLOv8, the model parameters have been reduced by 73.5 %, while accuracy, F1 score, and FPS have improved by 6.6 %, 4.3 %, and 116, respectively. Therefore, our proposed method is more lightweight and holds significant application value.

Currently, hydraulic fracturing technology is widely implemented for controlling the surrounding rock and enhancing permeability in low-permeability coal seams. Evaluating the effectiveness of hydraulic fracturing is a critical component of hydraulic fracturing operations. This study addresses the challenges in assessing the location, extension angle, and initial width of fracturing fractures within the current framework of hydraulic fracturing effectiveness evaluation in coal mines. We propose utilizing single-hole reflection borehole radar to evaluate the hydraulic fracturing effect, and through numerical simulation, we analyze the response characteristics of borehole radar when detecting various hydraulic fracturing-induced cracks. Initially, five models representing hydraulic fracturing cracks and two models for non-hydraulic fracturing cracks were established. Subsequently, the responses of borehole radar with central frequencies of 100, 200, and 400 MHz to cracks of identical shapes were analyzed. Additionally, the response characteristics of borehole radar with a 200 MHz central frequency to cracks of varying lengths (1, 2, 3 m), widths (4, 8, 40 cm), and angles (90°, 45°, 15°) were examined. Finally, a comparative analysis was conducted between hydraulic and non-hydraulic fracturing cracks. A branch hole was employed to simulate a hydraulic fracturing crack, allowing for an analysis of the borehole radar’s response characteristics in practical scenarios. The findings indicate that borehole radar is a viable tool for assessing hydraulic fracturing effects, providing a theoretical foundation for identifying the position of cracks, evaluating their effectiveness, and determining the regional effectiveness of the hydraulic fracturing crack system.

Deep learning (DL) has grown in popularity in civil inspection, notably for crack diagnosis, as a means of guaranteeing the long-term stability and security of concrete structures. It is critical to identify cracks to conduct inspections and assessments while preserving the existing concrete frameworks. This article reviews and analyses existing literature on identification of cracks on concrete structures using DL, to enhance the clarity and understanding of the ongoing research efforts in this domain. A systematic review found 97 linked research papers from 2018 to the beginning of 2023, using the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) Statement review process as a guide. The articles are categorised into several methods in identifying cracks, which include classification, segmentation, detection, and hybrid methods. Various issues in implementing DL in all the methods are discussed and several limitations, challenges and proposed solutions are presented. Finally, possible research directions are also discussed.

Due to continuous operations and manufacturing errors, fatigue cracks can emerge after hours of service; this causes a fan blade failure and potentially ruins an entire engine, turbo machinery, or rotating machinery of a similar kind. This paper focuses on the condition monitoring of the fan blade for detecting cracks occurring in these blades by analyzing the vibration responses. The mathematical formulation is carried out using the matrix method of dimensional analysis, which is dependent on the fundamental quantities of force, Length, Time, and Temperature (FLTƟ) systems of units. Numerical analysis in ANSYS software is done to comprehend the blade harmonic response for the cracked blade condition. Experimentation is also carried out on Tiera fault simulation machinery equipment, where vibration responses are measured and analyzed for crack detection in the blades. The tests were performed for three different cracks of different lengths and analyzed by varying parameters such as load speed, for which experiments are planned using a Box-Behnken design method. The test results were confirmed with the model equations developed, and notable similarities were seen between the analytical, numerical, and experimental analyses. Thus, the proposed study will help detect the cracks in the blades, thus reducing the serious accidents or failure of the machinery.

Rotational shafts are pivotal components in industrial settings and are responsible for transmitting torque and rotational motion. Despite their significance, these shafts are susceptible to faults, particularly cracks, which can adversely affect the system's performance and safety. Hence, efficient crack detection and diagnosis ensure safety, reliability, and costeffectiveness. This research aims to develop an Artificial Neural Network (ANN) model that can effectively identify cracks occurring at different depths and locations in rotating shafts, which operate at varying rotational speeds. Vibration signals were obtained and subjected to preprocessing using a bandpass filter to isolate the shaft signals from other components. Subsequently, time-domain statistical features were extracted from the filtered signals. An optimal feature selection methodology was employed to rank the extracted features, and the highest-ranking features were chosen for training the ANN model. The findings of this research indicate that the developed model achieved a classification accuracy of 94.4%.

Aiming at the problem that cracks with different shapes and scales on the exterior walls of high buildings are difficult to detect, this paper proposed a wall crack detection method for high-altitude cleaning robots by hybridizing the GAM attention mechanism and RepGFPN. First, the GAM attention mechanism was incorporated into the YOLOV5 backbone network to reduce information and amplify global features to improve the accuracy of feature extraction. Then, the neck network incorporated the RepFPN method to improve the descriptive ability of fused multi-scale features and to increase computational efficiency. Public datasets Concrete Crack Images for Classification, Mixed VOC2007, CrackForest-dataset-master, and UCMerced_LandUse were used for experimental validation. The ablation experiment results show that the average accuracy of mAP is improved by 13.5% after introducing the GAM attention mechanism under the yolov5 s original model, while the method in this paper (GR-YOLO) continues to improve by 4.7%. The experimental results show that the average accuracy mAP of the proposed method (GR-YOLO) is 24.0%, 47.1% and 41.0% higher than that of the model yolov5s + involution, yolov5s + p2 + involution and yolov5s + p2 + involution + CBAM, respectively. The method proposed in this article can more effectively improve the accuracy of crack detection and has important application prospects.

Industrial Computed Tomography (CT) serves as a pivotal non-destructive testing technique, significantly impacting the research and application domains of automatic defect detection in three-dimensional (3D) CT images. Recent advancements have showcased the exceptional capabilities of Convolutional Neural Networks (CNNs) in addressing image recognition and segmentation tasks. In the context of industrial 3D CT imaging, the paramount challenge lies in the segmentation of microstructural cracksâ task complicated by low grayscale contrast, small sizes, and imbalanced sample distributions. To tackle these complexities, we introduce a novel crack segmentation methodology grounded in multi-task deep convolutional neural networks.Our network architecture is composed of dual modules: a segmentation task module dedicated to crack extraction and a classification task module aimed at enhancing accuracy, with shared feature maps to leverage synergistic benefits. Inspired by the work of Chen et al., we incorporate the Efficient Channel Attention (ECA) block, an attention mechanism designed to refine segmentation accuracy. Empirical evaluations demonstrate that our proposed approach outperforms existing methodologies, showcasing superior performance in crack segmentation tasks. Notably, our network exhibits commendable generalization abilities. Despite being trained exclusively on simulated images, it yields robust and accurate predictions when applied to real-world CT images, indicating its adaptability across varied datasets. This study not only advances the technological frontier in industrial CT image analysis but also furnishes a potent tool and reference for future research endeavors in related fields.

Crack inspection is a routine practice in maintaining and monitoring civil infrastructure. Traditional inspection methods are time-consuming, raise safety-concerning, and impose significant cost on stakeholders. Fast and accurate concrete crack detection represents a cutting-edge technology and is an essential component for automated maintenance and monitoring features. Furthermore, lightweight networks are playing a crucial role in advancing the application of automation technologies. Thus, this paper proposes a lightweight embedded U-Net (LEU-Net) that emphasizes the trade-off between high accuracy and fast inference, aiming to achieve pixel-level segmentation of concrete crack. Motivated by the superior performance of IS-Net in the dichotomous image segmentation task, LEU-Net inherits its backbone. In order to efficiently reduce computational complexity and redundancy of convolutional kernels, DWConv and SCConv replace partial normal convolution within the architecture. To enhance the extraction of crack features across diverse dimensions, a GHPA (Group multi-axis Hadamard Product Attention mechanism) module based on Hadamard Product Attention mechanism is adopted in LEU-Net. During the training process, a hybrid loss function is employed to optimize the model training and enhance generalization ability. Our experimental results demonstrate that the proposed LEU-Net can deploy in a laptop with 6GB of VRAM and achieve 62.66% and 79.38% MIoU in the bridge crack image dataset and concrete pavement crack image dataset respectively, outperforming the other state-of-the-art segmentation networks and attaining more rapid and more accuracy crack segmentation.

Cracks are the most prevalent form of damage on pavement surfaces. Accurately recognizing pavement cracks is often difficult due to background interference and other challenges. Moreover, accurate and fast automatic detection of road cracks plays a crucial role in assessing pavement conditions. Therefore, a highly efficient lightweight network with only 0.78M parameters is proposed for the pixel-level pavement crack detection task. In this paper, adaptive enhancement module (AEM) is designed and added to the encoder network in order to avoid the problem of insufficient model learning capability due to the use of depth-wise separable convolutions. Meanwhile, a coordinate-aware fusion module (CFM) is proposed, which fully fuses skip connection features and decoder features to enhance cross-channel interaction information. Comprehensive experimental results demonstrate that the proposed network outperforms existing methods across four public datasets: CamCrack789, CFD, DeepCrack237, and Crack500, achieving F1 scores of 94.6%, 92.8%, 91.0%, and 79.8%, respectively. Furthermore, ablation study confirmed the efficacy of both the AEM and the CFM.

Computer vision which uses Convolutional Neural Network (CNN) models is a robust and accurate tool for precise monitoring and pixel-level detection of potential damage in concrete structures. Using a state-of-the-art Dense ResU-Net model integrated with T-Max-Avg pooling layers, the present study introduces a novel and effective method for crack detection in concrete structures. The major innovation of this research is the introduction of the T-Max-Avg pooling layer within the Dense ResU-Net architecture which synergistically combines the strengths of both max and average pooling to improve feature retention and minimize information loss during crack detection. In addition, the incorporation of Residual and Dense blocks within the U-Net framework significantly enhances feature extraction and network depth, resulting in a more robust anomaly detection. The implementation of extensive data augmentation techniques improves the robustness of the model while the application of spatial dropout and L2 regularization techniques prevents overfitting. The proposed model showed a superior performance, outperforming traditional and state-of-the-art models. It had a Dice Coefficient score of 97.41%, an Intersection-over-Union (IoU) score of 98.63%, and an accuracy of 99.2% using a batch size of 32. These results confirmed the reliability and efficacy of the Dense ResU-Net with T-Max-Avg pooling layer for accurate crack detection, demonstrating its potential for real-world applications in structural health monitoring. By taking advantage of advanced deep learning techniques, the proposed method addressed the limitations of traditional crack detection techniques and offered significant improvements in robustness and accuracy.

The efficient and accurate detection of cracks in concrete structures is critical for maintaining structural integrity and safety. This study compares two state-of-the-art convolutional neural network (CNN) models, Mask R-CNN and YOLOv8, for automated concrete crack detection, each model representing two mainstream approaches for object detection and instance segmentation: single-stage and two-stage approach. We evaluate both models on 1,203 concrete images with 7:2:1 training, testing, and validation split, and assess their accuracy and processing speed. Mask R-CNN achieves a mean Intersection over Union (IoU) of 96.5% with a minimum IoU of 77% and higher consistency, compared to YOLOv8’s 90.6%, which often shows complete failure with IoU of 0%. In terms of computation speed, YOLOv8 shows 0.3225 s of average processing time per image, slightly outperforming the speed of Mask R-CNN, 0.4867 s. Despite YOLOv8’s faster processing speed, considering the characteristics of concrete crack detection tasks where accuracy should be prioritized over speed, Mask R-CNN seems a more proper model for reliable crack detection. We also show the accuracy of Mask R-CNN for crack detection tasks can be further enhanced by employing the ResNeXt backbone.

Detection of cracks in solar photovoltaic (PV) modules is crucial for optimal performance and long-term reliability. The development of convolutional neural networks (CNNs) has significantly improved crack detection, offering improved accuracy and efficiency over traditional methods. This paper presents a comprehensive review and comparative analysis of CNN-based approaches for crack detection in solar PV modules. The review discusses various CNN architectures, including custom-designed networks and pre-trained models, as well as data-augmentation techniques and ensemble learning methods. Additionally, challenges related to limited dataset sizes, generalizability across different solar panels, interpretability of CNN models, and real-time detection are discussed. The review also identifies opportunities for future research, such as the need for larger and more diverse datasets, model interpretability, and optimized computational speed. Overall, this paper serves as a valuable resource for researchers and practitioners interested in using CNNs for crack detection in solar PV modules.

Abstract This paper proposes a weakly-supervised structural surface crack detection algorithm that can detect the crack area in an image with low data labeling cost. The algorithm consists of a convolutional neural networks Vgg16-Crack for classification, an improved and optimized class activation map (CAM) algorithm for accurately reflecting the position and distribution of cracks in the image, and a method that combines superpixel segmentation algorithm simple linear iterative clustering (SLIC) with CAM for more accurate semantic segmentation of cracks. In addition, this paper uses Bayesian optimization algorithm to obtain the optimal parameter combination that maximizes the performance of the model. The test results show that the algorithm only requires image-level labeling, which can effectively reduce the labor and material consumption brought by pixel-level labeling while ensuring accuracy.

In recent times, the deployment of advanced structural health monitoring techniques has increased due to the aging infrastructural elements. This paper employed an enhanced You Only Look Once (YOLO) v4-tiny algorithm, based on the Crack Detection Model (CDM), to accurately identify and classify crack types in reinforced concrete (RC) members. YOLOv4-tiny is faster and more efficient than its predecessors, offering real-time detection with reduced computational complexity. Despite its smaller size, it maintains competitive accuracy, making it ideal for applications requiring high-speed processing on resource-limited devices. First, an extensive experimental program was conducted by testing full-scale RC members under different shear span (a) to depth ratios to achieve flexural and shear dominant failure modes. The digital images captured from the failure of RC beams were analyzed using the CDM of the YOLOv4-tiny algorithm. Results reveal the accurate identification of cracks formed along the depth of the beam at different stages of loading. Moreover, the confidence score attained for all the test samples was more than 95%, which indicates the accuracy of the developed model in capturing the types of cracks in the RC beam. The outcomes of the proposed work encourage the use of a developed CDM algorithm in real-time crack detection analysis of critical infrastructural elements.

Cracks pose a critical challenge in the preservation of historical buildings worldwide, particularly in fair-faced walls, where timely and accurate detection is essential to prevent further degradation. Traditional image processing methods have proven inadequate for effectively detecting building cracks. Despite global advancements in deep learning, crack detection under diverse environmental and lighting conditions remains a significant technical hurdle, as highlighted by recent international studies. To address this challenge, we propose an enhanced crack detection algorithm, CL-YOLOv8 (ConvNeXt V2-LSKA-YOLOv8). By integrating the well-established ConvNeXt V2 model as the backbone network into YOLOv8, the algorithm benefits from advanced feature extraction techniques, leading to a superior detection accuracy. This choice leverages ConvNeXt V2’s recognized strengths, providing a robust foundation for improving the overall model performance. Additionally, by introducing the LSKA (Large Separable Kernel Attention) mechanism into the SPPF structure, the feature receptive field is enlarged and feature correlations are strengthened, further enhancing crack detection accuracy in diverse environments. This study also contributes to the field by significantly expanding the dataset for fair-faced wall crack detection, increasing its size sevenfold through data augmentation and the inclusion of additional data. Our experimental results demonstrate that CL-YOLOv8 outperforms mainstream algorithms such as Faster R-CNN, YOLOv5s, YOLOv7-tiny, SSD, and various YOLOv8n/s/m/l/x models. CL-YOLOv8 achieves an accuracy of 85.3%, a recall rate of 83.2%, and a mean average precision (mAP) of 83.7%. Compared to the YOLOv8n base model, CL-YOLOv8 shows improvements of 0.9%, 2.3%, and 3.9% in accuracy, recall rate, and mAP, respectively. These results underscore the effectiveness and superiority of CL-YOLOv8 in crack detection, positioning it as a valuable tool in the global effort to preserve architectural heritage.

In the field of construction engineering, the cracking of concrete structures is a common engineering problem, which has a great impact on the overall stability and service life of the engineered structure. During structural repair, crack detection is the most critical step. Automatic detection significantly reduces the engineering cost and human factor error compared with manual detection. However, due to the changeable environment of the project site and different image specifications, using a single algorithm makes it difficult to balance high efficiency and high accuracy. In this study, we designed a combined recognition method including the region growth algorithm and machine learning regression that can achieve a tradeoff between accuracy and efficiency. Firstly, the regression method learns the image features of the dataset and the specific region growth threshold, and the regression function is trained by using the open-source dataset to determine the region growth threshold using the characteristics of the images included in the tests. The region growth algorithm is used to expand the threshold from the seed points of the image to obtain the crack recognition results. The results show that this method improves the accuracy of SSIM by 7% compared with the traditional region growth algorithm, and does not significantly increase the computational cost, with an increase of 0.78 s per photo process. Compared with the deep learning method, the recognition accuracy of SSIM is decreased by 5.96%, but it takes less resources and has high efficiency.