Search

The conventional method for detecting road defects relies heavily on manual inspections, which are often inefficient and struggle with precise defect localization. This paper introduces a novel approach for identifying and locating road defects based on an enhanced ML-YOLO algorithm. By refining the YOLOv8 object detection framework, we optimize both the traditional convolutional layers and the spatial pyramid pooling network. Additionally, we incorporate the Convolutional Block Attention to effectively capture channel and spatial features, along with the Selective Kernel Networks that dynamically adapt to feature extraction across varying scales. An optimized target localization algorithm is proposed to achieve high-precision identification and accurate positioning of road defects. Experimental results indicate that the detection accuracy of the improved ML-YOLO algorithm reaches 0.841, with a recall rate of 0.745 and an average precision of 0.817. Compared to the baseline YOLOv8 model, there is an increase in accuracy by 0.13, a rise in recall rate by 0.117, and an enhancement in average precision by 0.116. After the high detection accuracy of road defects was confirmed, generalization experiments were carried out on the improved ML-YOLO model in the public data set. The experimental results showed that compared with the original YOLOv8n, the average precision and recall rate of all types of ML-YOLO increased by 0.075, 0.121, and 0.035 respectively, indicating robust generalization capabilities. When applied to real-time road monitoring scenarios, this algorithm facilitates precise detection and localization of defects while significantly mitigating traffic accident risks and extending roadway service life. A high detection accuracy of road defects was achieved. [ABSTRACT FROM AUTHOR]

Due to the progress in light detection and ranging (LiDAR) technology, the collection of road point cloud data containing depth information and spatial coordinates has become more accessible. Consequently, utilizing point cloud data for pavement distress detection and quantification emerges as a crucial approach to improving the precision and reliability of road maintenance procedures. This paper aims to automatically detect and visualize pavement distress using LiDAR, deep learning-based 3D object detection method, and building information modeling (BIM). A pavement distress data set is first established using the point cloud data obtained from LiDAR. Then, the 3D object detection network, namely PointPillar, is employed for pavement distress detection, and the detection results will be quantified at a region-level. Finally, pavement BIM model integrating parametrically modeled distress families is built to visually manage the detected distress. After training and validating the model with the pavement distress data set, a detection performance index of recall is 78.5%, mean average precision (mAP) is 62.7%, which is better than other compared point cloud-based methods though the detection performance can be further improved. In addition, a newly untrained section of road is applied for the experiment. The detected distress is integrated in BIM environment for a visual management, providing a better maintenance guidance. [ABSTRACT FROM AUTHOR]

Nowadays, road maintenance is a crucial planning task for all road authorities across the world, making them spend vast amounts of money on identification and rehabilitation programmes every year. Thus, many researchers, road engineers, and decision-makers have turned to a system called pavement management system, the critical step of which is pavement inspection. So far, several methods have been proposed for pavement distress data collection and evaluation. These methods have also evolved with the advancement of technology and intelligent transportation systems. Today, researchers have acknowledged autonomous vehicles as a sophisticated tool monitoring right of way for appropriate operating; however, little attention has been paid to using collected pavement condition data for pavement management. This study reviews the technologies and algorithms proposed so far to investigate the feasibility of using the data regularly collected by these vehicles to evaluate pavement conditions. We hope our paper paves the way for further research. [ABSTRACT FROM AUTHOR]

With the development of state-of-the-art algorithms, pavement distress can already be detected automatically. However, most pavement distress detection is currently implemented as a single task, either at the region level or at the pixel level. To comprehensively assess the pavement condition, a multitask fusion model, Pavement Distress Detection Network (PDDNet), was proposed for integrated pavement distress detection at both the region level and pixel level. PDDNet was trained and tested on distress images captured via unmanned aerial vehicle (UAV), and seven types of pavement distresses were investigated and analyzed. Compared with Mask Region-based Convolutional Neural Network (R-CNN), U-Net, and W-segnet, PDDNet shows higher performance in classification, localization, and segmentation of pavement distresses. Results demonstrate that PDDNet achieves region-level and pixel-level detection of seven types of distresses with the mean average precision of 0.810 and 0.795, respectively. As a portable and lightweight device, the UAV can collect full-width pavement distress images, which helps improve the efficiency of pavement distress detection. [ABSTRACT FROM AUTHOR]

Pavement distress detection is crucial in road health assessment and monitoring. However, there are still some challenges in extracting pavement distress based on deep learning: such as insufficient segmentation, extraction errors and discontinuities. In this paper, we propose DARNet, a network for pavement distress extraction. A Distress Aware Attention Module (DAAM) is proposed to solve the problem of discontinuity in distress extraction due to inaccurate recovery of distress pixels during upsampling. Based on the characteristics of distress morphology, a Refinement Extraction Module (REM) is designed to effectively capture horizontal and vertical pavement damage features by fusing high-level and low-level features, which improves the accuracy of the model in extracting details of pavement damage information. Finally, a Weighted Cross-Entropy Loss function (WCEL) is introduced to assign weights according to the distance of the pixel point to the boundary of the distress, which solves the problem that the traditional cross entropy function treats each pixel point equally. We also propose a set of pavement distress datasets LNTU_RDD_GS, and the experimental results show that DARNet can reach 82.68% mIoU and 90.13% F score in the datasets in this paper, 80.63% mIoU and 88.35% F score in the four public datasets.

Research on road infrastructure structural health monitoring is critical due to the increasing problem of deteriorated conditions. The traditional approach to pavement distress detection relies on human-based visual recognition, a time-consuming and labor-intensive method. While Deep Learning-based computer vision systems are the most promising approach, they face the challenges of reduced performance due to the scarcity of labeled data due, high annotation costs misaligned with engineering applications, and limited instances of minority defects. This paper introduces a novel generative diffusion model for data augmentation, creating synthetic images of rare defects. It also investigates methods to enhance image quality and reduce production time. Compared to Generative Adversarial Networks, the optimal configuration excels in reliability, quality and diversity. After incorporating synthetic images into the training of our pavement distress detector, YOLOv5, its mean average precision has been enhanced. This computer-aided system enhances recognition and labelling efficiency, promoting intelligent maintenance and repairs.

The UAV-PDD2023 dataset consists of pavement distress images captured by unmanned aerial vehicles (UAVs) in China with more than 11,150 instances under two different weather conditions and across varying levels of construction quality. The roads in the dataset consist of highways, provincial roads, and county roads constructed under different requirements. It contains six typical types of pavement distress instances, including longitudinal cracks, transverse cracks, oblique cracks, alligator cracks, patching, and potholes. The dataset can be used to train deep learning models for automatically detecting and classifying pavement distresses using UAV images. In addition, the dataset can be used as a benchmark to evaluate the performance of different algorithms for solving tasks such as object detection, image classification, etc. The UAV-PDD2023 dataset can be downloaded for free at the URL in this paper.

Copyright of Journal of Polytechnic is the property of Journal of Polytechnic and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Pavement distress data in a single section usually presents a long-tailed distribution, with potholes, sealed cracks, and other distresses normally located at the tail. This distribution will seriously affect the performance and robustness of big data-driven deep learning detection models. Conventional data augmentation algorithms only expand the amount of data by image transformation and fail to enlarge the data diversity. Due to such a drawback, this paper proposes a novel two-stage pavement distress image augmentation pattern, in which a mask is generated randomly according to the geometric features of the distress in the first stage; and in the second stage, a distress-free pavement image with the fused mask is transformed into a pavement distress image. Furthermore, two convolutional networks, M-DCGAN and MDTMN, are designed to complete the generation task in two stages separately. In comparison with other generation algorithms, the quality and diversity of the generation results of proposed algorithms are better than other algorithms. In addition, distress detection tests are conducted which indicate that the expanded dataset can raise the IoU from 48.83% to 83.65% at maximum, and the augmented data by the proposed algorithm contributes more to the detection performance. [ABSTRACT FROM AUTHOR]

Surging vehicle loads and changing climate environments place significant stress on road infrastructure. Pavement management requires fast and effective methods of detecting pavement distress and perform timely maintenance. This paper presents in detail the hardware devices for automated data collection and the 2D and 3D image acquisition methods. The detection methods for different pavement distresses are comprehensively analyzed and summarized in the review. In addition, the review covers the latest and classical artificial intelligence (AI) image processing algorithms, including traditional image processing, machine learning, and deep learning methods applied in pavement distress detection. The review summarizes the challenges, limitations, emerging technologies, and future trends of AI algorithms. The review findings indicate that the application of AI technology methods in pavement distress detection has grown dramatically, but challenges still exist in AI technology application in practical engineering. [Display omitted] • The review covers the key hardware used in automated data collection vehicles. • This paper reviews the use of artificial intelligence for detecting various pavement distresses. • This article reviews pavement distress detection methods from a knowledge graph perspective. • This paper collates the application of deep learning algorithms in pavement distress recognition. [ABSTRACT FROM AUTHOR]

Pavement distress detection (PDD) plays a vital role in planning timely pavement maintenance that improves pavement service life. In order to promote the development of PDD technologies and find out the insufficiencies in PDD field, this paper reviews the technical development history and characteristics of various PDD technologies, which contributes to the current state of research on PDD. First, processes of PDD are briefly introduced. The PDD technologies based on radar ranging, 2D image, laser ranging and 3D structured light are illustrated. The newest 3D PDD technology based on interference fringe, which has better accuracy, is in progress. The principles and implementation processes of these methods are discussed. Finally, the shortcomings of these technologies in the field of PDD are concluded. Recommendations for future development are provided. The research results show that various PDD technologies have been continuously improved, developed, over the past decade, and have achieved a series of results. However, the measurements from existing PDD technologies can not be metrological traced to acquire the true dimensions of pavement distresses. The lack of metrological traceability technology in the PDD field needs to be further solved. In order to achieve more accurate and efficient PDD, the metrological traceability technology of PDD systems has become the future development direction in this field.

Copyright of Journal of Southeast University / Dongnan Daxue Xuebao is the property of Journal of Southeast University Editorial Office and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

A comprehensive representation of the road pavement state of health is of great interest. In recent years, automated data collection and processing technology has been used for pavement inspection. In this paper, a new signal on graph (SoG) model of road pavement distresses is presented with the aim of improving automatic pavement distress detection systems. A novel nonlinear Bayesian estimator in recovering distress metrics is also derived. The performance of the methodology was evaluated on a large dataset of pavement distress values collected in field tests conducted in Kazakhstan. The application of the proposed methodology is effective in recovering acquisition errors, improving road failure detection. Moreover, the output of the Bayesian estimator can be used to identify sections where the measurement acquired by the 3D laser technology is unreliable. Therefore, the presented model could be used to schedule road section maintenance in a better way. [ABSTRACT FROM AUTHOR]

Timely detection of early pavement distress is crucial for effective pavement maintenance. A promising approach involves the segmenting the road surface cracks area from its images. However, the previously employed full-resolution residual network (FRRN), when employing a large number of full-resolution residual units (FRRUs), is susceptible to challenges such as exploding and vanishing gradients. Addressing these issues, this paper has introduced a depth supervision mechanism into FRRN, leading to the development of the Depth Supervision FRRN (DSFRRN) model, designed specifically for accurate segmentation of cracks in road images. This model enhances nonlinear expressiveness while retaining essential features, achieving high precision in pavement crack segmentation. Through experimental validation using a dataset comprising 1565 images of damaged pavements, the result demonstrate that DSFRRN outperforms the original FRRN in terms of both F1 score and Jaccard index, accurately delineating cracks. Consequently, the automated pavement damage detection method presented herein facilitates cost-effective and efficient pavement inspection, contributing to the safeguarding of roadway surfaces and transportation infrastructure. [ABSTRACT FROM AUTHOR]

Pavement distress detection is of significance for road maintenance and traffic safety. Manual pavement distress detection suffers from high workloads, inefficiency, low accuracy, and high cost. To replace the manual operations in the pre-filling detection with the aim to improve efficiency and reduce cost, this paper proposes a three-stage automatic inspection and evaluation system for pavement distress based on improved deep convolutional neural networks (CNNs). First, the system integrates multi-level context information from the CNN classification model to construct discriminative super-features to determine whether there is distress in the pavement image and the type of the distress, so as to achieve rapid detection of pavement distress. Then, the pavement images with distress are fed into the CNN segmentation model to highlight the distress region with pixel-wise. In the segmentation model, a novel pyramid feature extraction module and a novel guidance attention mechanism are introduced. Finally, we evaluate the degree of pavement damage according to the segmentation results of the CNN segmentation model. In the experiments, we compare our classification model and segmentation model with other state-of-the-art methods on two pavement distress datasets, and the results demonstrate that the proposed models achieve out-performance on different evaluation metrics. [ABSTRACT FROM AUTHOR]

We present a novel deep learning framework named the Iteratively Optimized Patch Label Inference Network (IOPLIN) for automatically detecting various pavement distresses that are not solely limited to specific ones, such as cracks and potholes. IOPLIN can be iteratively trained with only the image label via the Expectation-Maximization Inspired Patch Label Distillation (EMIPLD) strategy, and accomplish this task well by inferring the labels of patches from the pavement images. IOPLIN enjoys many desirable properties over the state-of-the-art single branch CNN models such as GoogLeNet and EfficientNet. It is able to handle images in different resolutions, and sufficiently utilize image information particularly for the high-resolution ones, since IOPLIN extracts the visual features from unrevised image patches instead of the resized entire image. Moreover, it can roughly localize the pavement distress without using any prior localization information in the training phase. In order to better evaluate the effectiveness of our method in practice, we construct a large-scale Bituminous Pavement Disease Detection dataset named CQU-BPDD consisting of 60,059 high-resolution pavement images, which are acquired from different areas at different times. Extensive results on this dataset demonstrate the superiority of IOPLIN over the state-of-the-art image classification approaches in automatic pavement distress detection. The source codes of IOPLIN are released on https://github.com/DearCaat/ioplin , and the CQU-BPDD dataset is able to be accessed on https://dearcaat.github.io/CQU-BPDD/. [ABSTRACT FROM AUTHOR]

Pavement distress detection is of great significance for road maintenance and to ensure road safety. At present, detection methods based on deep learning have achieved outstanding performance in related fields. However, these methods require large-scale training samples. For pavement distress detection, it is difficult to collect more images with pavement distress, and the types of pavement diseases are increasing with time, so it is impossible to ensure sufficient pavement distress samples to train the supervised deep model. In this article, we propose a new few-shot pavement distress detection method based on metric learning, which can effectively learn new categories from a few labeled samples. In this article, we adopt the backend network (ResNet18) to extract multilevel feature information from the base classes and then send the extracted features into the metric module. In the metric module, we introduce the attention mechanism to learn the feature attributes of “what” and “where” and focus the model on the desired characteristics. We also introduce a new metric loss function to maximize the distance between different categories while minimizing the distance between the same categories. In the testing stage, we calculate the cosine similarity between the support set and query set to complete novel category detection. The experimental results show that the proposed method significantly outperforms several benchmarking methods on the pavement distress dataset (the classification accuracies of 5-way 1-shot and 5-way 5-shot are 77.20% and 87.28%, respectively). [ABSTRACT FROM AUTHOR]

• A lightweight feature fusion network called CFPN is proposed to enhance pavement distress detection efficiency in complex environments. • The CFPN facilitates direct interactions between non-adjacent levels to comprehend context and scale information related to pavement distresses. • A new pavement distress detection model is proposed by integrating CFPN with WIoUv2 into the YOLOv5s framework. • The proposed model achieves higher accuracy and inference speed with lower parameters than compared models. Efficiently detecting pavement distress in complex environments is crucial for the intelligent operation of transportation infrastructure. This study proposed a novel pavement distress detection model based on You Only Look Once version 5 (YOLOv5) incorporating a novel lightweight feature fusion network named crossed feature pyramid network (CFPN) and an improved loss function to enhance pavement distress detection efficiency in complex environments. The proposed model was evaluated by a dataset comprising 7076 images representing four common pavement distress classes. The experimental results indicate the proposed model outperforms in challenging working conditions such as shadows and overlapped multi-object bounding boxes. The proposed model achieves mean average precision (mAP), recall, precision, and frames per second (FPS) values of 69.3 %, 65.7 %, 73.3 %, and 118, respectively. These values are 4.0 %, 0.7 %, 4.2 %, and 9.3 % higher than those of YOLOv5s, but the parameters are squeezed by 27.1 %, expanding its application in non-destructive automatic pavement distress detection. [ABSTRACT FROM AUTHOR]

Automated pavement distress detection benefits road maintenance and operation by providing the condition and location of various distress rapidly. Existing work generally relies on manual labor or specific algorithms trained by dedicated datasets, which hinders the efficiency and applicable scenarios of methods. Street view map provides interactive panoramas of a large scale of urban roadway network, and is updated in a recurrent manner by the provider. This paper proposed a deep learning method based on a pre-trained neural network architecture to identify and locate different distress in real-time. About 20,000 street view images were collected and labeled as the training dataset using the Baidu e-map. Eight types of distress are notated using Yolov3 deep learning architecture. The scale-invariant feature transform (SIFT) descriptors combined with GPS and bounding boxes were applied to judge the deterioration of the distress. A decision tree was designed to evaluate the change of the distress over some time. A typical road in Shanghai was selected to verify the effectiveness of the proposed model. The images of the road from 2015 to 2017 were collected from the street view map. The results showed that the mean average precision of the proposed algorithm is 88.37%, demonstrating the vast potential of applying this method to detect pavement distress. 43 distress were newly generated, and 49 previous distress were patched in the two years. The proposed method can assist the authorities to schedule the maintenance activities more effectively.

The condition of municipal roads has deteriorated considerably in recent years, leading to large scale pavement distress such as cracks or potholes. In order to enable road maintenance, pavement distress should be timely detected. However, manual investigation, which is still the most widely applied approach toward pavement assessment, puts maintenance personnel at risk and is time-consuming. During the last decade, several efforts have been made to automatically assess the condition of the municipal roads without any human intervention. Vehicles are equipped with sensors and cameras in order to collect data related to pavement distress and record videos of the pavement surface. Yet, this data are usually not processed while driving, but instead it is recorded and later analyzed off-line. As a result, a vast amount of memory is required to store the data and the available memory may not be sufficient. To reduce the amount of saved data, the authors have previously proposed a graphics processing units (GPU)-enabled pavement distress detection approach based on the wavelet transform of pavement images. The GPU implementation enables pavement distress detection in real time. Although the method used in the approach provides very good results, the method can still be improved by incorporating pavement surface texture characteristics. This paper presents an implementation of textural features on GPUs for pavement distress detection. Textural features are based on gray-tone spatial dependencies in an image and characterize the image texture. To evaluate the computational efficiency of the GPU implementation, performance tests are carried out. The results show that the speedup achieved by implementing the textural features on the GPU is sufficient to enable real-time detection of pavement distress. In addition, classification results obtained by applying the approach on 16,601 pavement images are compared to the results without integrating textural features. There results demonstrate that an improvement of 27% is achieved by incorporating pavement surface texture characteristics. [ABSTRACT FROM AUTHOR]

The performance of deep learning-based computer vision systems for road infrastructure assessment is hindered by the scarcity of real-world, high-volume public datasets. Current research predominantly focuses on crack detection and segmentation, without devising end-to-end systems capable of effectively evaluating the most affected roads and assessing the out-of-sample performance. To address these limitations, this study proposes a public dataset with annotations of 7099 images and 13 types of defects, not only based on cracks, for the confrontation and development of deep learning models. These images are used to train and compare YOLOv5 sub-models based on pure detection efficiency, and standard object detection metrics, to select the optimum architecture. A novel post-processing filtering mechanism is then designed, which reduces the false positive detections by 20.5%. Additionally, a pavement condition index (ASPDI) is engineered for deep learning-based models to identify areas in need for immediate maintenance. To facilitate decision-making by road administrations, a software application is created, which integrates the ASPDI, geotagged images, and detections. This tool has allowed to detect two road sections in critical need of repair. The refined architecture is validated on open datasets, achieving mean average precision scores of 0.563 and 0.570 for RDD2022 and CPRI, respectively. • Open-source road distress dataset design with object detection annotations. • Comparative analysis of Deep Learning YOLOv5 models for pavement distress detection. • Novel rule-based post-processing with 20.5% reduction in false positives. • Development of road condition index and quantification of most damaged areas. • Validation of the best post-processed model across various open-source datasets. [ABSTRACT FROM AUTHOR]

Efficient road maintenance is imperative for infrastructure longevity and safety. Conventional ground vehicle-based methods for detecting pavement distress, however, encounter limitations in practice when dealing with complex road structures. Drones, endowed with greater spatial freedom, can access road segments that are hard-to-reach to ground vehicles, thereby enhancing detection efficiency and expanding detection coverage. By harnessing the complementary strengths of both detection modalities, we propose a scheme that capitalizes on the cooperative coordination of drones and ground vehicles for effective pavement distress detection. Our proposed scheme is evaluated using realistic road networks in practice. Results reveal that the coordinated detection scheme strikes a favorable balance between fixed device-related expenses and detection efficiency. This scheme offers promising policy implications, streamlining maintenance across diverse road networks and meeting extensive infrastructure needs, offering policymakers an efficient and viable scheme for road infrastructure maintenance. [ABSTRACT FROM AUTHOR]

Copyright of Technisches Messen is the property of De Gruyter and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Pavements function as protective layers for roads and require frequent inspection and maintenance throughout their service life. This paper describes an intelligent pavement distress inspection system that uses an enhanced version of the 'you only look once' (YOLO) model and an omni-scale network (OSNet) to instantly capture road surface distress images and their precise locations. The YOLO model was evaluated on a dataset comprising 9749 pavement distress images, with the detected distress serving as an input for feature extraction and instance-level recognition through OSNet. The OSNet model achieved a mean average precision (mAP) of 99.4% for a dataset containing 398 individual distress instances. The proposed methods were successfully integrated into a pavement distress inspection vehicle. Field experiments demonstrated the real-time capability and high efficiency of the system, with significant improvement in road maintenance inspection efficiency • An intelligent pavement disease inspection system that utilizes an improved version of YOLO and OSNet is proposed. • The space-to-depth strategy was introduced in the backbone of the YOLOv5s model. • OSNet was developed for feature extraction and filter out repeated distress images. • The proposed methods were integrated into an intelligent pavement disease inspection vehicle. • Field experiments confirmed the real-time and high efficiency of the system. [ABSTRACT FROM AUTHOR]

This paper proposes a novel algorithm called three-dimensional (3D) shadow modeling for the detection of various descended patterns on 3D pavement surfaces to improve detection accuracy; these patterns include pavement cracks, potholes, joints, and grooves. Analogous to a filtering algorithm that only preserves the useful signals, the proposed 3D shadow modeling is a general-purpose algorithm that can transform the raw 3D pavement data into a binary domain in which the amount of unneeded data, such as texture variations and noises in the original 3D data, can be reduced effectively and the remaining descended patterns then become distinctive. Bidirectional lighting is the essential concept specifically used in the 3D shadow modeling to find shadowed areas that are lower than the local surroundings. With the projection angle changes from 0 to 90°, the proposed 3D shadow modeling can generate diverse solutions from being extremely sensitive to extremely insensitive. In other words, the proposed 3D shadow modeling can satisfy varying needs on the basis of sensitivity. In complement to the detection of various descended patterns on pavement surface, the proposed 3D shadow modeling can be employed to detect ascended patterns once the original 3D pavement surface is vertically inverted. The experimental results demonstrated that the proposed 3D shadow modeling is an efficient algorithm and can yield a high level of precision (92.37%) and recall (92.93%) for the detection of typical descended patterns (potholes, cracks, joints, and grooves) on all selected 3D pavement images. [ABSTRACT FROM AUTHOR]

Automated pavement distress detection based on 2D images is facing various challenges. To efficiently complete the crack and pothole segmentation in a practical environment, an automated pixel-level pavement distress detection framework integrating stereo vision and deep learning is developed in this study. Based on the multi-view stereo imaging system, multi-feature pavement image datasets containing color images, depth images and color-depth overlapped images are established, providing a new perspective for deep learning. To alleviate computational burden, a modified U-net deep learning architecture introducing depthwise separable convolution is proposed for crack and pothole segmentation. These methods are tested in asphalt roads with different circumstances. The results show that the 3D pavement image achieves millimeter-level accuracy. The enhanced 3D crack segmentation model outperforms other models in terms of segmentation accuracy and inference speed. After obtaining the high-resolution pothole segmentation map, the automated pothole volume measurement is realized with high accuracy. • Stereo vision and deep learning were integrated for automated pavement crack and pothole segmentation. • Multi-feature image datasets containing 2D, 3D and enhanced-3D images were established by stereo vision. • A modified U-net embedding depthwise separable convolution was proposed for faster segmentation. • The deep learning efficiency using different types of images was investigated. • Automated pothole volume measurement was achieved based on 3D image segmentation. [ABSTRACT FROM AUTHOR]

• A study using GPR signals and NINs for pavement distress detection is presented. • The NIN-based method detected pavement distresses with high precision. • The NIN stability was not affected by GPR transmitting frequencies. • The NIN-based method had a distinct superiority in detection effectiveness. This study proposes a nondestructive testing technique for pavement distress detection using ground-penetrating radar and network in networks. Ground-penetrating radar signals are imported into two network-in-network structure as input data directly. The network in networks are used as deep learning models to distinguish abnormal signals, recognize distress types, and measure distress locations and sizes. A database with information from four highways is generated by a ground-penetrating radar with different transmitting frequencies and numbers of samples per trace. Then, the database is used to train, validate, and test the network in networks. The results show that the proposed method detects cracks, water-damage pits, and uneven settlements with 85.17% accuracy, 2.15 mm location errors, and reasonable stability. The proposed method was superior to other state-of-the-art techniques in terms of classification accuracy, location error, and stability. Additionally, the results show that this method overcomes the negative effect of transmitting frequencies in pavement distress detection using GPR data. [ABSTRACT FROM AUTHOR]