•  
  •  
 

Corresponding Author

Kamel Mahdy

Subject Area

Civil and Environmental Engineering

Article Type

Review

Abstract

For the last few decades, researchers have been devising a simple and cost-effective method to evaluate pavement distresses to give decision-makers adequate feedbacks about the pavement condition of a certain road. Fortunately, with the evolution and progression of computer vision tools and techniques, good results had been achieved regarding the detection, classification, and quantification of road distress. In this paper, a new efficient process of road distress analysis using deep learning models is introduced. This new process was tested on a collected road dataset to evaluate the efficiency and speed of this low-cost road maintenance system. Promising results were obtained from the proposed process based on the deep learning model used with an outstanding performance of ~400 fps and distress detection every ~5 cm for a vehicle moving at 40 km/h. Furthermore, the output of the developed process was used as an input for the Pavement Condition Index (PCI) calculation module to determine the pavement condition of the road on a single-day mission. The proposed system focuses on detecting some specific types of distresses: Alligator cracks, longitudinal cracks, transverse cracks, block cracks, lane longitudinal cracks, reflective cracks, and sealed cracks. Experimental results show that this process based on deep learning models achieved promising results of ~5% difference from the true PCI, currently calculated in a month, just in a single day using very low-cost methods.

Keywords

Pavement, Pavement Management System (PMS), Pavement Distresses, Pavement maintenance Management System (PMMS), Pavement Condition Index (PCI), Deep Learning. NASNET

Creative Commons License

Creative Commons Attribution 4.0 License
This work is licensed under a Creative Commons Attribution 4.0 License.

Download

Share

COinS