Deep Neural Networks for Visual Bridge Inspections and Defect Visualisation in Civil Engineering

Julia Bush; Tadeo Corradi; Jelena Ninic; Georgia Thermou

doi:10.14279/depositonce-12021

Deep Neural Networks for Visual Bridge Inspections and Defect Visualisation in Civil Engineering

Julia Bush
, Tadeo Corradi
, Jelena Ninic
, Georgia Thermou

Civil Engineering

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

1454 Downloads (Pure)

Abstract

Ageing infrastructure is a global concern, and current structural health monitoring practices are coming under review. With a view to streamline the visual bridge inspection process, we assess the classification performance of two Deep Neural Networks, VGG16 and MobileNet, on a challenging dataset of over 70,000 unprocessed bridge inspection images of three defect categories: corrosion, crack, and spalling. Grad-CAM “heatmap” visualisations on VGG16 predictions provide a coarse localisation of the defect region and some insight into the functioning of the network. Similar performance is attained on MobileNet, for applications where speed or computational cost is a consideration. We conclude that with further optimisation this approach could have an application in automated defect tagging.

Original language	English
Title of host publication	EG-ICE 2021 Workshop on Intelligent Computing in Engineering
Editors	Jimmy Abualdenien, André Borrmann, Lucian-Constantin Ungureanu, Timo Hartmann
Place of Publication	Berlin
Publisher	Universitätsverlag der Technischen Universität Berlin
Pages	421–431
Number of pages	11
ISBN (Electronic)	9783798332126
ISBN (Print)	9783798332119
DOIs	https://doi.org/10.14279/depositonce-12021
Publication status	Published - 3 Jan 2022
Event	EG-ICE 2021 Workshop on Intelligent Computing in Engineering - Hybrid, Technische Universität Berlin, Berlin, Germany Duration: 30 Jun 2021 → 2 Jul 2021

Workshop

Workshop	EG-ICE 2021 Workshop on Intelligent Computing in Engineering
Country/Territory	Germany
City	Berlin
Period	30/06/21 → 2/07/21

Access to Document

10.14279/depositonce-12021Licence: Creative Commons: Attribution (CC BY)

EG-ICE_2021Final published version, 36.5 MBLicence: Creative Commons: Attribution (CC BY)

https://library.oapen.org/handle/20.500.12657/52189Licence: Creative Commons: Attribution (CC BY)

Cite this

Bush, J., Corradi, T., Ninic, J., & Thermou, G. (2022). Deep Neural Networks for Visual Bridge Inspections and Defect Visualisation in Civil Engineering. In J. Abualdenien, A. Borrmann, L.-C. Ungureanu, & T. Hartmann (Eds.), EG-ICE 2021 Workshop on Intelligent Computing in Engineering (pp. 421–431). Universitätsverlag der Technischen Universität Berlin. https://doi.org/10.14279/depositonce-12021

Bush, Julia ; Corradi, Tadeo ; Ninic, Jelena et al. / Deep Neural Networks for Visual Bridge Inspections and Defect Visualisation in Civil Engineering. EG-ICE 2021 Workshop on Intelligent Computing in Engineering. editor / Jimmy Abualdenien ; André Borrmann ; Lucian-Constantin Ungureanu ; Timo Hartmann . Berlin : Universitätsverlag der Technischen Universität Berlin, 2022. pp. 421–431

@inproceedings{617c77558cb14cefaaf7cbfa97dfa381,

title = "Deep Neural Networks for Visual Bridge Inspections and Defect Visualisation in Civil Engineering",

abstract = "Ageing infrastructure is a global concern, and current structural health monitoring practices are coming under review. With a view to streamline the visual bridge inspection process, we assess the classification performance of two Deep Neural Networks, VGG16 and MobileNet, on a challenging dataset of over 70,000 unprocessed bridge inspection images of three defect categories: corrosion, crack, and spalling. Grad-CAM “heatmap” visualisations on VGG16 predictions provide a coarse localisation of the defect region and some insight into the functioning of the network. Similar performance is attained on MobileNet, for applications where speed or computational cost is a consideration. We conclude that with further optimisation this approach could have an application in automated defect tagging.",

author = "Julia Bush and Tadeo Corradi and Jelena Ninic and Georgia Thermou",

year = "2022",

month = jan,

day = "3",

doi = "10.14279/depositonce-12021",

language = "English",

isbn = "9783798332119",

pages = "421–431",

editor = "Jimmy Abualdenien and Andr{\'e} Borrmann and Lucian-Constantin Ungureanu and \{Hartmann \}, Timo",

booktitle = "EG-ICE 2021 Workshop on Intelligent Computing in Engineering",

publisher = "Universit{\"a}tsverlag der Technischen Universit{\"a}t Berlin",

note = "EG-ICE 2021 Workshop on Intelligent Computing in Engineering ; Conference date: 30-06-2021 Through 02-07-2021",

}

Bush, J, Corradi, T, Ninic, J & Thermou, G 2022, Deep Neural Networks for Visual Bridge Inspections and Defect Visualisation in Civil Engineering. in J Abualdenien, A Borrmann, L-C Ungureanu & T Hartmann (eds), EG-ICE 2021 Workshop on Intelligent Computing in Engineering. Universitätsverlag der Technischen Universität Berlin, Berlin, pp. 421–431, EG-ICE 2021 Workshop on Intelligent Computing in Engineering, Berlin, Brandenburg, Germany, 30/06/21. https://doi.org/10.14279/depositonce-12021

Deep Neural Networks for Visual Bridge Inspections and Defect Visualisation in Civil Engineering. / Bush, Julia; Corradi, Tadeo; Ninic, Jelena et al.
EG-ICE 2021 Workshop on Intelligent Computing in Engineering. ed. / Jimmy Abualdenien; André Borrmann; Lucian-Constantin Ungureanu; Timo Hartmann . Berlin: Universitätsverlag der Technischen Universität Berlin, 2022. p. 421–431.

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

TY - GEN

T1 - Deep Neural Networks for Visual Bridge Inspections and Defect Visualisation in Civil Engineering

AU - Bush, Julia

AU - Corradi, Tadeo

AU - Ninic, Jelena

AU - Thermou, Georgia

PY - 2022/1/3

Y1 - 2022/1/3

N2 - Ageing infrastructure is a global concern, and current structural health monitoring practices are coming under review. With a view to streamline the visual bridge inspection process, we assess the classification performance of two Deep Neural Networks, VGG16 and MobileNet, on a challenging dataset of over 70,000 unprocessed bridge inspection images of three defect categories: corrosion, crack, and spalling. Grad-CAM “heatmap” visualisations on VGG16 predictions provide a coarse localisation of the defect region and some insight into the functioning of the network. Similar performance is attained on MobileNet, for applications where speed or computational cost is a consideration. We conclude that with further optimisation this approach could have an application in automated defect tagging.

AB - Ageing infrastructure is a global concern, and current structural health monitoring practices are coming under review. With a view to streamline the visual bridge inspection process, we assess the classification performance of two Deep Neural Networks, VGG16 and MobileNet, on a challenging dataset of over 70,000 unprocessed bridge inspection images of three defect categories: corrosion, crack, and spalling. Grad-CAM “heatmap” visualisations on VGG16 predictions provide a coarse localisation of the defect region and some insight into the functioning of the network. Similar performance is attained on MobileNet, for applications where speed or computational cost is a consideration. We conclude that with further optimisation this approach could have an application in automated defect tagging.

U2 - 10.14279/depositonce-12021

DO - 10.14279/depositonce-12021

M3 - Conference contribution

SN - 9783798332119

SP - 421

EP - 431

BT - EG-ICE 2021 Workshop on Intelligent Computing in Engineering

A2 - Abualdenien, Jimmy

A2 - Borrmann, André

A2 - Ungureanu, Lucian-Constantin

A2 - Hartmann , Timo

PB - Universitätsverlag der Technischen Universität Berlin

CY - Berlin

T2 - EG-ICE 2021 Workshop on Intelligent Computing in Engineering

Y2 - 30 June 2021 through 2 July 2021

ER -

Bush J, Corradi T, Ninic J, Thermou G. Deep Neural Networks for Visual Bridge Inspections and Defect Visualisation in Civil Engineering. In Abualdenien J, Borrmann A, Ungureanu LC, Hartmann T, editors, EG-ICE 2021 Workshop on Intelligent Computing in Engineering. Berlin: Universitätsverlag der Technischen Universität Berlin. 2022. p. 421–431 Epub 2021 Aug 6. doi: 10.14279/depositonce-12021

Deep Neural Networks for Visual Bridge Inspections and Defect Visualisation in Civil Engineering

Abstract

Workshop

Access to Document

Fingerprint

Cite this