FasteNet: A Fast Railway Fastener Detector

Jun Jet Tai; Mauro Innocente; Owais Mehmood

doi:10.1007/978-981-16-2377-6_71

FasteNet: A Fast Railway Fastener Detector

Jun Jet Tai, Mauro Innocente, Owais Mehmood

Centre for Future Transport and Cities

Omnicom Balfour Beatty

Research output: Chapter in Book/Report/Conference proceeding › Conference proceeding › peer-review

2 Citations (Scopus)

60 Downloads (Pure)

Abstract

In this work, a novel high-speed railway fastener detector is introduced. This fully convolutional network, dubbed FasteNet, foregoes the notion of bounding boxes and performs detection directly on a predicted saliency map. FasteNet uses transposed convolutions and skip connections, the effective receptive field of the network is 1.5 times larger than the average size of a fastener, enabling the network to make predictions with high confidence, without sacrificing output resolution. In addition, due to the saliency map approach, the network is able to vote for the presence of a fastener up to 30 times per fastener, boosting prediction accuracy. FasteNet is capable of running at 110 FPS on an Nvidia GTX 1080, while taking in inputs of 1600 x 512 with an average of 14 fasteners per image. Our source is open here: https://github.com/jjshoots/DL FasteNet.git.

Original language	English
Title of host publication	Proceedings of 6th International Congress on Information and Communication Technology, ICICT 2021
Subtitle of host publication	ICICT 2021, London, Volume 1
Editors	Xin-She Yang, Simon Sherratt, Nilanjan Dey, Amit Joshi
Publisher	Springer Nature
Pages	767-777
Number of pages	11
ISBN (Electronic)	978-981-16-2102-4
ISBN (Print)	978-981-16-2101-7
DOIs	https://doi.org/10.1007/978-981-16-2377-6_71
Publication status	Published - Jan 2022
Event	6th International Congress on Information and Communication Technology - Online, London, United Kingdom Duration: 25 Feb 2021 → 26 Feb 2021 Conference number: 6 https://www.icict.co.uk/

Publication series

Name	Lecture Notes in Networks and Systems
Volume	235
ISSN (Print)	2367-3370
ISSN (Electronic)	2367-3389

Conference

Conference	6th International Congress on Information and Communication Technology
Abbreviated title	ICICT 2021
Country/Territory	United Kingdom
City	London
Period	25/02/21 → 26/02/21
Internet address	https://www.icict.co.uk/

Bibliographical note

Copyright © and Moral Rights are retained by the author(s) and/ or other copyright owners. A copy can be downloaded for personal non-commercial research or study, without prior permission or charge. This item cannot be reproduced or quoted extensively from without first obtaining permission in writing from the copyright holder(s). The content must not be changed in any way or sold commercially in any format or medium without the formal permission of the copyright holders.

Keywords

Convolutional neural network
computer vision
railway fastener detector
Object detection
Convolutional neural networks
Railway fastener detection

ASJC Scopus subject areas

Signal Processing
Control and Systems Engineering
Computer Networks and Communications

Access to Document

10.1007/978-981-16-2377-6_71

Post-PrintAccepted author manuscript, 15.1 MB

Cite this

Tai, J. J., Innocente, M., & Mehmood, O. (2022). FasteNet: A Fast Railway Fastener Detector. In X.-S. Yang, S. Sherratt, N. Dey, & A. Joshi (Eds.), Proceedings of 6th International Congress on Information and Communication Technology, ICICT 2021: ICICT 2021, London, Volume 1 (pp. 767-777). (Lecture Notes in Networks and Systems; Vol. 235). Springer Nature. https://doi.org/10.1007/978-981-16-2377-6_71

FasteNet: A Fast Railway Fastener Detector. / Tai, Jun Jet ; Innocente, Mauro; Mehmood, Owais.
Proceedings of 6th International Congress on Information and Communication Technology, ICICT 2021: ICICT 2021, London, Volume 1. ed. / Xin-She Yang; Simon Sherratt; Nilanjan Dey; Amit Joshi. Springer Nature, 2022. p. 767-777 (Lecture Notes in Networks and Systems; Vol. 235).

Research output: Chapter in Book/Report/Conference proceeding › Conference proceeding › peer-review

Tai, JJ , Innocente, M & Mehmood, O 2022, FasteNet: A Fast Railway Fastener Detector. in X-S Yang, S Sherratt, N Dey & A Joshi (eds), Proceedings of 6th International Congress on Information and Communication Technology, ICICT 2021: ICICT 2021, London, Volume 1. Lecture Notes in Networks and Systems, vol. 235, Springer Nature, pp. 767-777, 6th International Congress on Information and Communication Technology, London, United Kingdom, 25/02/21. https://doi.org/10.1007/978-981-16-2377-6_71

Tai JJ , Innocente M, Mehmood O. FasteNet: A Fast Railway Fastener Detector. In Yang XS, Sherratt S, Dey N, Joshi A, editors, Proceedings of 6th International Congress on Information and Communication Technology, ICICT 2021: ICICT 2021, London, Volume 1. Springer Nature. 2022. p. 767-777. (Lecture Notes in Networks and Systems). Epub 2021 Sept 24. doi: 10.1007/978-981-16-2377-6_71

@inproceedings{43e23acf0bbb4aa2b32af82ba0810374,

title = "FasteNet: A Fast Railway Fastener Detector",

abstract = "In this work, a novel high-speed railway fastener detector is introduced. This fully convolutional network, dubbed FasteNet, foregoes the notion of bounding boxes and performs detection directly on a predicted saliency map. FasteNet uses transposed convolutions and skip connections, the effective receptive field of the network is 1.5 times larger than the average size of a fastener, enabling the network to make predictions with high confidence, without sacrificing output resolution. In addition, due to the saliency map approach, the network is able to vote for the presence of a fastener up to 30 times per fastener, boosting prediction accuracy. FasteNet is capable of running at 110 FPS on an Nvidia GTX 1080, while taking in inputs of 1600 x 512 with an average of 14 fasteners per image. Our source is open here: https://github.com/jjshoots/DL FasteNet.git.",

keywords = "Convolutional neural network, computer vision, railway fastener detector, Object detection, Convolutional neural networks, Railway fastener detection",

author = "Tai, {Jun Jet} and Mauro Innocente and Owais Mehmood",

note = "Copyright {\textcopyright} and Moral Rights are retained by the author(s) and/ or other copyright owners. A copy can be downloaded for personal non-commercial research or study, without prior permission or charge. This item cannot be reproduced or quoted extensively from without first obtaining permission in writing from the copyright holder(s). The content must not be changed in any way or sold commercially in any format or medium without the formal permission of the copyright holders. ; 6th International Congress on Information and Communication Technology, ICICT 2021 ; Conference date: 25-02-2021 Through 26-02-2021",

year = "2022",

month = jan,

doi = "10.1007/978-981-16-2377-6_71",

language = "English",

isbn = "978-981-16-2101-7",

series = "Lecture Notes in Networks and Systems",

publisher = "Springer Nature",

pages = "767--777",

editor = "Xin-She Yang and Simon Sherratt and Nilanjan Dey and Amit Joshi",

booktitle = "Proceedings of 6th International Congress on Information and Communication Technology, ICICT 2021",

address = "United Kingdom",

url = "https://www.icict.co.uk/",

}

TY - GEN

T1 - FasteNet: A Fast Railway Fastener Detector

AU - Tai, Jun Jet

AU - Innocente, Mauro

AU - Mehmood, Owais

N1 - Conference code: 6

PY - 2022/1

Y1 - 2022/1

N2 - In this work, a novel high-speed railway fastener detector is introduced. This fully convolutional network, dubbed FasteNet, foregoes the notion of bounding boxes and performs detection directly on a predicted saliency map. FasteNet uses transposed convolutions and skip connections, the effective receptive field of the network is 1.5 times larger than the average size of a fastener, enabling the network to make predictions with high confidence, without sacrificing output resolution. In addition, due to the saliency map approach, the network is able to vote for the presence of a fastener up to 30 times per fastener, boosting prediction accuracy. FasteNet is capable of running at 110 FPS on an Nvidia GTX 1080, while taking in inputs of 1600 x 512 with an average of 14 fasteners per image. Our source is open here: https://github.com/jjshoots/DL FasteNet.git.

AB - In this work, a novel high-speed railway fastener detector is introduced. This fully convolutional network, dubbed FasteNet, foregoes the notion of bounding boxes and performs detection directly on a predicted saliency map. FasteNet uses transposed convolutions and skip connections, the effective receptive field of the network is 1.5 times larger than the average size of a fastener, enabling the network to make predictions with high confidence, without sacrificing output resolution. In addition, due to the saliency map approach, the network is able to vote for the presence of a fastener up to 30 times per fastener, boosting prediction accuracy. FasteNet is capable of running at 110 FPS on an Nvidia GTX 1080, while taking in inputs of 1600 x 512 with an average of 14 fasteners per image. Our source is open here: https://github.com/jjshoots/DL FasteNet.git.

KW - Convolutional neural network

KW - computer vision

KW - railway fastener detector

KW - Object detection

KW - Convolutional neural networks

KW - Railway fastener detection

UR - http://www.scopus.com/inward/record.url?scp=85116424900&partnerID=8YFLogxK

U2 - 10.1007/978-981-16-2377-6_71

DO - 10.1007/978-981-16-2377-6_71

M3 - Conference proceeding

SN - 978-981-16-2101-7

T3 - Lecture Notes in Networks and Systems

SP - 767

EP - 777

BT - Proceedings of 6th International Congress on Information and Communication Technology, ICICT 2021

A2 - Yang, Xin-She

A2 - Sherratt, Simon

A2 - Dey, Nilanjan

A2 - Joshi, Amit

PB - Springer Nature

T2 - 6th International Congress on Information and Communication Technology

Y2 - 25 February 2021 through 26 February 2021

ER -

FasteNet: A Fast Railway Fastener Detector

Abstract

Publication series

Conference

Bibliographical note

Keywords

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this