Automated classification and detection of multiple pavement distress images based on deep learning

To achieve automatic,fast,efficient and high-precision pavement distress classification and detection,road surface distress image classification and detection models based on deep learning are trained.First,a pavement distress image dataset is built,including 9017pictures with distress,and 9620 pictures without distress.These pictures were captured from 4 asphalt highways of 3 provinces in China.In each pavement distress image,there exists one or more types of distress,including alligator crack,longitudinal crack,block crack,transverse crack,pothole and patch.The distresses are labeled by a rectangle bounding box on the pictures.Then ResNet networks and VGG networks are used respectively as binary classification models for distressed and non-distressed imagines classification,and as multi-label classification models for six types of distress classification.Training techniques,such as data augmentation,batch normalization,dropout,momentum,weight decay,transfer learning,and discriminative learning rate are used in training the model.Among the 4 CNNs considered in this study,namely ResNet 34 and 50,and VGG 16 and 19,for the binary classification,ResNet 50 has the highest Accuracy of 96.243%,Precision of 95.183%,and ResNet 34 has the highest Recall of 97.824%,and F2 score of 97.052%.For multi-label classification,ResNet 50 has the best performance,with the highest Accuracy of 90.257%,higher than 90%required by the Chinese standard(JTG H20-2018)for road distresses detection,F2 score-82.231%,and Precision-76.509%,and ResNet34 has the highest Recall of 87.32%.To locate and quantify the distress areas in the images,the single shot multibox detector(SSD)model is developed,in which the ResNet 50 is used as the base network to extract features.When the intersection over union(IoU)is set to 0,0.25,0.50,0.75,the mean average precision(mAP)of the model are found to be 74.881%,50.511%,28.432%,3.969%,respectively.

Vision and Geolocation Data Combination for Precise Human Detection and Tracking in Search and Rescue Operations

In this paper, a study and evaluation of the combination of GPS/GNSS techniques and advanced image processing algorithms for distressed human detection, positioning and tracking, from a fully autonomous Unmanned Aerial Vehicle (UAV)-based rescue support system, are presented. In particular, the issue of human detection both on terrestrial and marine environment under several illumination and background conditions, as the human silhouette in water differs significantly from a terrestrial one, is addressed. A robust approach, including an adaptive distressed human detection algorithm running every N input image frames combined with a much faster tracking algorithm, is proposed. Real time or near-real-time distressed human detection rates achieved, using a single, low cost day/night NIR camera mounted onboard a fully autonomous UAV for Search and Rescue (SAR) operations. Moreover, the generation of our own dataset, for the image processing algorithms training is also presented. Details about both hardware and software configuration as well as the assessment of the proposed approach performance are fully discussed. Last, a comparison of the proposed approach to other human detection methods used in the literature is presented.

A review on pavement distress and structural defects detection and quantification technologies using imaging approaches

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 3 D 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.

Influence of computation algorithm on the accuracy of rut depth measurement

Rutting is one of the dominant pavement distresses, hence, the accuracy of rut depth measurements can have a substantial impact on the maintenance and rehabilitation （M 8： R） strategies and funding allocation. Different computation algorithms such as straight- edge method and wire line method, which are based on the same raw data, may lead to rut depth estimation which are not always consistent. Therefore, there is an urgent need to assess the impact of algorithm types on the accuracy of rut depth computation. In this paper, a 1B-point-based laser sensor detection technology, commonly accepted in China for rut depth measurements, was used to obtain a database of 85,000 field transverse profiles having three representative rutting shapes with small, medium and high severity rut levels. Based on the reconstruction of real transverse profiles, the consequences from two different algorithms were compared. Results showed that there is a combined effect of rut depth and profile shape on the rut depth computation accuracy. As expected, the dif- ference between the results obtained with the two computation methods increases with deeper rutting sections： when the distress is above 15 mm （severe level）, the average dif- ference between the two computation methods is above 1.5 mm, normally, the wire line method provides larger results. The computation suggests that the rutting shapes have a minimal influence on the results. An in-depth analysis showed that the upheaval outside of the wheel path is a dominant shape factor which results in higher computation differences.