Research On Pavement Crack Inspection And Total Quality Management Based On Deep Learning

With the development of the pavement construction,the requirements for the quality of pavements are rising.The quality of the pavement plays an important role in the economic structure sustainable development.However,due to the increasing service time,the long-term load,the nature disaster as well as the man-made damage,the damages of pavement are increasing,which leads to the occurrence of security incidents.Therefore,in order to extend the service life of the road and ensure its safety,we must have a regular inspection and detect the damage part of the pavement structure.Then,maintenance measures are utilized to extend the service life of it.This can also prevent disasters happening.Currently,manual inspection mainly constitutes the crack detection of the pavement structure.This approach has a strong subjectivity and a high error rate.Besides,it is time-consuming and dangerous.The crack detection technology of the pavement based on image recognition technology has solved these problems and becomes very popular in this research area.This paper proposed an automatic inspection and positioning system(Auto-DIT)for the pavement crack detection by using deep learning methods.The proposed Auto-DIT system utilizes advanced UAV to gather the images of the pavement and displays the detection results to terminal users by developing the Android application.It can quickly obtain the image information of the pavement and detect the defect types automatically.At the same time,the position of the defect within the image can be framed and the geographical position of the image can be localized.Finally,all results can be clearly shown to workers through the mobile terminals.Additionally,this paper analyzes the existing problems in the current quality management of the pavement structure and a new comprehensive quality management method based on Auto-DIT system is proposed.The automatic inspection and localization system(Auto-DIT)of the pavement crack detection by utilizing deep learning method is proposed by this paper,which can quickly collect the pavement images,accurately detect and locate the pavement cracks.It improves the pavement crack detection effectively and can be used in the real applications.In addition,through the analysis of engineering quality management theories and the study of existing problems in pavement quality management,this paper presents a new pavement quality management based on Auto-DIT system,which not only enriches the theoretical research in the field of pavement quality management,but also can guide the practice of road quality management,thus it can ensure the life cycle quality of pavement better.