A Deep Learning based Process Model for Crack Detection in Pavement Structures

No Thumbnail Available

Authors

Rababaah, Aaron

Issue Date

2022-03-22

Type

Conference Papers funded by AUK

Language

Keywords

Research Projects

Organizational Units

Journal Issue

Alternative Title

Abstract

This paper presents an investigation of the effectiveness of a Machine Deep Learning (DL) model using Convolution Neural Networks (CNN) in detecting cracks in asphalt pavement. Pavement cracks have been studied for decades using traditional Machine Learning (ML) methods such as neural networks, genetic algorithms, fuzzy logic, etc. In this work, we present a DL model based on CNN to study the effectiveness of modern machine vision methods on an old problem of pavement crack detection. We provide our methodology in developing the proposed model and the validation process in this paper. A dataset consisting of 500 sample images was used to test the model and our experiments showed that the proposed model is effective with a mean accuracy of 96% and a standard deviation of. 025. Future work is recommended to be on crack type classification after the successful detection process.

Description

Citation

Rababaah, A. (23-25MAR, 2022). A Deep Learning based Process Model for Crack Detection in PavementStructures. In 2022 9th International Conference on Computing for Sustainable GlobalDevelopment (INDIACom); IEEE Conference ID: 51348, (pp. 1-6). doi:10.23919/INDIACom54597.2022.9763286

Publisher

IEEE Conference

License

Journal

Volume

Issue

PubMed ID

DOI

ISSN

EISSN