Sequential Score Adaptation with Extreme Value Theory for Robust Railway Track Inspection

Xavier Gibert; Vishal M. Patel; Rama Chellappa

doi:10.1109/ICCVW.2015.27

Sequential Score Adaptation with Extreme Value Theory for Robust Railway Track Inspection

Xavier Gibert, Vishal M. Patel, Rama Chellappa

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

Abstract

Periodic inspections are necessary to keep railroad tracks in state of good repair and prevent train accidents. Automatic track inspection using machine vision technology has become a very effective inspection tool. Because of its non-contact nature, this technology can be deployed on virtually any railway vehicle to continuously survey the tracks and send exception reports to track maintenance personnel. However, as appearance and imaging conditions vary, false alarm rates can dramatically change, making it difficult to select a good operating point. In this paper, we use extreme value theory (EVT) within a Bayesian framework to optimally adjust the sensitivity of anomaly detectors. We show that by approximating the lower tail of the probability density function (PDF) of the scores with an Exponential distribution (a special case of the Generalized Pareto distribution), and using the Gamma conjugate prior learned from the training data, it is possible to reduce the variability in false alarm rate and improve the overall performance. This method has shown an increase in the defect detection rate of rail fasteners in the presence of clutter (at PFA 0.1%) from 95.40% to 99.26% on the 85-mile Northeast Corridor (NEC) 2012-2013 concrete tie dataset.

Original language	English (US)
Title of host publication	Proceedings - 2015 IEEE International Conference on Computer Vision Workshops, ICCVW 2015
Publisher	Institute of Electrical and Electronics Engineers Inc.
Pages	131-138
Number of pages	8
ISBN (Electronic)	9781467383905
DOIs	https://doi.org/10.1109/ICCVW.2015.27
State	Published - Feb 11 2016
Externally published	Yes
Event	15th IEEE International Conference on Computer Vision Workshops, ICCVW 2015 - Santiago, Chile Duration: Dec 11 2015 → Dec 18 2015

Publication series

Name	Proceedings of the IEEE International Conference on Computer Vision
Volume	2015-February
ISSN (Print)	1550-5499

Conference

Conference	15th IEEE International Conference on Computer Vision Workshops, ICCVW 2015
Country/Territory	Chile
City	Santiago
Period	12/11/15 → 12/18/15

Keywords

Bayes methods
Detectors
Fasteners
Feature extraction
Indexes
Inspection
Robustness

ASJC Scopus subject areas

Software
Computer Vision and Pattern Recognition

Access to Document

10.1109/ICCVW.2015.27

Cite this

Gibert, X., Patel, V. M., & Chellappa, R. (2016). Sequential Score Adaptation with Extreme Value Theory for Robust Railway Track Inspection. In Proceedings - 2015 IEEE International Conference on Computer Vision Workshops, ICCVW 2015 (pp. 131-138). Article 7406376 (Proceedings of the IEEE International Conference on Computer Vision; Vol. 2015-February). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/ICCVW.2015.27

Sequential Score Adaptation with Extreme Value Theory for Robust Railway Track Inspection. / Gibert, Xavier; Patel, Vishal M.; Chellappa, Rama.
Proceedings - 2015 IEEE International Conference on Computer Vision Workshops, ICCVW 2015. Institute of Electrical and Electronics Engineers Inc., 2016. p. 131-138 7406376 (Proceedings of the IEEE International Conference on Computer Vision; Vol. 2015-February).

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

Gibert, X, Patel, VM & Chellappa, R 2016, Sequential Score Adaptation with Extreme Value Theory for Robust Railway Track Inspection. in Proceedings - 2015 IEEE International Conference on Computer Vision Workshops, ICCVW 2015., 7406376, Proceedings of the IEEE International Conference on Computer Vision, vol. 2015-February, Institute of Electrical and Electronics Engineers Inc., pp. 131-138, 15th IEEE International Conference on Computer Vision Workshops, ICCVW 2015, Santiago, Chile, 12/11/15. https://doi.org/10.1109/ICCVW.2015.27

Gibert X, Patel VM, Chellappa R. Sequential Score Adaptation with Extreme Value Theory for Robust Railway Track Inspection. In Proceedings - 2015 IEEE International Conference on Computer Vision Workshops, ICCVW 2015. Institute of Electrical and Electronics Engineers Inc. 2016. p. 131-138. 7406376. (Proceedings of the IEEE International Conference on Computer Vision). doi: 10.1109/ICCVW.2015.27

Gibert, Xavier ; Patel, Vishal M. ; Chellappa, Rama. / Sequential Score Adaptation with Extreme Value Theory for Robust Railway Track Inspection. Proceedings - 2015 IEEE International Conference on Computer Vision Workshops, ICCVW 2015. Institute of Electrical and Electronics Engineers Inc., 2016. pp. 131-138 (Proceedings of the IEEE International Conference on Computer Vision).

@inproceedings{af733309800549c19b01438e67f4d7a4,

title = "Sequential Score Adaptation with Extreme Value Theory for Robust Railway Track Inspection",

abstract = "Periodic inspections are necessary to keep railroad tracks in state of good repair and prevent train accidents. Automatic track inspection using machine vision technology has become a very effective inspection tool. Because of its non-contact nature, this technology can be deployed on virtually any railway vehicle to continuously survey the tracks and send exception reports to track maintenance personnel. However, as appearance and imaging conditions vary, false alarm rates can dramatically change, making it difficult to select a good operating point. In this paper, we use extreme value theory (EVT) within a Bayesian framework to optimally adjust the sensitivity of anomaly detectors. We show that by approximating the lower tail of the probability density function (PDF) of the scores with an Exponential distribution (a special case of the Generalized Pareto distribution), and using the Gamma conjugate prior learned from the training data, it is possible to reduce the variability in false alarm rate and improve the overall performance. This method has shown an increase in the defect detection rate of rail fasteners in the presence of clutter (at PFA 0.1%) from 95.40% to 99.26% on the 85-mile Northeast Corridor (NEC) 2012-2013 concrete tie dataset.",

keywords = "Bayes methods, Detectors, Fasteners, Feature extraction, Indexes, Inspection, Robustness",

author = "Xavier Gibert and Patel, {Vishal M.} and Rama Chellappa",

note = "Publisher Copyright: {\textcopyright} 2015 IEEE.; 15th IEEE International Conference on Computer Vision Workshops, ICCVW 2015 ; Conference date: 11-12-2015 Through 18-12-2015",

year = "2016",

month = feb,

day = "11",

doi = "10.1109/ICCVW.2015.27",

language = "English (US)",

series = "Proceedings of the IEEE International Conference on Computer Vision",

publisher = "Institute of Electrical and Electronics Engineers Inc.",

pages = "131--138",

booktitle = "Proceedings - 2015 IEEE International Conference on Computer Vision Workshops, ICCVW 2015",

}

TY - GEN

T1 - Sequential Score Adaptation with Extreme Value Theory for Robust Railway Track Inspection

AU - Gibert, Xavier

AU - Patel, Vishal M.

AU - Chellappa, Rama

PY - 2016/2/11

Y1 - 2016/2/11

N2 - Periodic inspections are necessary to keep railroad tracks in state of good repair and prevent train accidents. Automatic track inspection using machine vision technology has become a very effective inspection tool. Because of its non-contact nature, this technology can be deployed on virtually any railway vehicle to continuously survey the tracks and send exception reports to track maintenance personnel. However, as appearance and imaging conditions vary, false alarm rates can dramatically change, making it difficult to select a good operating point. In this paper, we use extreme value theory (EVT) within a Bayesian framework to optimally adjust the sensitivity of anomaly detectors. We show that by approximating the lower tail of the probability density function (PDF) of the scores with an Exponential distribution (a special case of the Generalized Pareto distribution), and using the Gamma conjugate prior learned from the training data, it is possible to reduce the variability in false alarm rate and improve the overall performance. This method has shown an increase in the defect detection rate of rail fasteners in the presence of clutter (at PFA 0.1%) from 95.40% to 99.26% on the 85-mile Northeast Corridor (NEC) 2012-2013 concrete tie dataset.

AB - Periodic inspections are necessary to keep railroad tracks in state of good repair and prevent train accidents. Automatic track inspection using machine vision technology has become a very effective inspection tool. Because of its non-contact nature, this technology can be deployed on virtually any railway vehicle to continuously survey the tracks and send exception reports to track maintenance personnel. However, as appearance and imaging conditions vary, false alarm rates can dramatically change, making it difficult to select a good operating point. In this paper, we use extreme value theory (EVT) within a Bayesian framework to optimally adjust the sensitivity of anomaly detectors. We show that by approximating the lower tail of the probability density function (PDF) of the scores with an Exponential distribution (a special case of the Generalized Pareto distribution), and using the Gamma conjugate prior learned from the training data, it is possible to reduce the variability in false alarm rate and improve the overall performance. This method has shown an increase in the defect detection rate of rail fasteners in the presence of clutter (at PFA 0.1%) from 95.40% to 99.26% on the 85-mile Northeast Corridor (NEC) 2012-2013 concrete tie dataset.

KW - Bayes methods

KW - Detectors

KW - Fasteners

KW - Feature extraction

KW - Indexes

KW - Inspection

KW - Robustness

UR - http://www.scopus.com/inward/record.url?scp=84962040460&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84962040460&partnerID=8YFLogxK

U2 - 10.1109/ICCVW.2015.27

DO - 10.1109/ICCVW.2015.27

M3 - Conference contribution

AN - SCOPUS:84962040460

T3 - Proceedings of the IEEE International Conference on Computer Vision

SP - 131

EP - 138

BT - Proceedings - 2015 IEEE International Conference on Computer Vision Workshops, ICCVW 2015

PB - Institute of Electrical and Electronics Engineers Inc.

T2 - 15th IEEE International Conference on Computer Vision Workshops, ICCVW 2015

Y2 - 11 December 2015 through 18 December 2015

ER -

Sequential Score Adaptation with Extreme Value Theory for Robust Railway Track Inspection

Abstract

Publication series

Conference

Keywords

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this