A platform for research: civil engineering, architecture and urbanism
A platform for research: civil engineering, architecture and urbanism
Pavement Pothole Monitoring via Artificial Intelligence Technology
In recent decades, the rapid development of highway construction in various countries (especially highways) has greatly promoted the development of regional economies. After decades of rapid development, highways have become an important part of road transportation in all countries. An artificial intelligence algorithms have been prevalently utilized for 3D road imaging and pothole monitoring for over two decades. In this work, a novel artificial intelligence algorithm for road imaging and pothole monitoring is proposed. This algorithm is prepared to increase the level of pothole automation when combined with inspection vehicles. The developed pothole monitoring model is based on deep recurrent neural network (RNN) model with filter rolling around an input volume and generating an output instead of feedforward neural network (FFNN) since the pavement pothole propagation is a time dependent and memory dependent behaviour. For the dataset used in this paper, modelled with high accuracy rate is 92.56%, regression rate is 90.63%, and F-score is 87.42%.
Pavement Pothole Monitoring via Artificial Intelligence Technology
In recent decades, the rapid development of highway construction in various countries (especially highways) has greatly promoted the development of regional economies. After decades of rapid development, highways have become an important part of road transportation in all countries. An artificial intelligence algorithms have been prevalently utilized for 3D road imaging and pothole monitoring for over two decades. In this work, a novel artificial intelligence algorithm for road imaging and pothole monitoring is proposed. This algorithm is prepared to increase the level of pothole automation when combined with inspection vehicles. The developed pothole monitoring model is based on deep recurrent neural network (RNN) model with filter rolling around an input volume and generating an output instead of feedforward neural network (FFNN) since the pavement pothole propagation is a time dependent and memory dependent behaviour. For the dataset used in this paper, modelled with high accuracy rate is 92.56%, regression rate is 90.63%, and F-score is 87.42%.
Pavement Pothole Monitoring via Artificial Intelligence Technology
Dangui, Guo (author) / Hong, Weixing (author) / Altabey, Wael A. (author)
2023-09-24
1159679 byte
Conference paper
Electronic Resource
English