Automatic annotation of subsea pipelines using deep learning

Research output: Contribution to journalArticle

22 Downloads (Pure)

Abstract

Regulatory requirements for sub-sea oil and gas operators mandates the frequent inspection of pipeline assets to ensure that their degradation and damage are maintained at acceptable levels. The inspection process is usually sub-contracted to surveyors who utilise sub-sea Remotely Operated Vehicles (ROVs), launched from a surface vessel and piloted over the pipeline. ROVs capture data from various sensors/instruments which are subsequently reviewed and interpreted by human operators, creating a log of event annotations; a slow, labour-intensive and costly process. The paper presents an automatic image annotation framework that identifies/classifies key events of interest in the video footage viz. exposure, burial, field joints, anodes and free spans. The reported methodology utilises transfer learning with a Deep Convolutional Neural Network (ResNet-50), fine-tuned on real-life, representative data from challenging sub-sea environments with low lighting conditions, sand agitation, sea-life and vegetation. The network outputs are configured to perform multi-label image classifications for the critical events. The annotation performance varies between 95.1% and 13 99.7% in terms of accuracy and 90.4% and 99.4% in terms of F1-Score depending on event type. The performance results are on a per-frame basis and corroborate the potential of the algorithm to be the foundation for an intelligent decision support framework that automates the annotation process. The solution can execute annotations in real-time and is significantly more cost-effective than human-only approaches.
Original languageEnglish
Article number674
Number of pages15
JournalSensors
Volume20
Issue number3
DOIs
Publication statusPublished - 26 Jan 2020

Fingerprint

annotations
Remotely operated vehicles
Oceans and Seas
learning
Pipelines
Inspection
Learning
Image classification
Labels
Data acquisition
Anodes
Oils
Sand
Lighting
Gases
Personnel
inspection
Neural networks
vehicles
Burial

Keywords

  • visual inspection
  • subsea pipeline survey
  • multi-label image classification
  • deep learning
  • transfer learning

Cite this

@article{d2ae80ac0dcf486c957576bd074e2942,
title = "Automatic annotation of subsea pipelines using deep learning",
abstract = "Regulatory requirements for sub-sea oil and gas operators mandates the frequent inspection of pipeline assets to ensure that their degradation and damage are maintained at acceptable levels. The inspection process is usually sub-contracted to surveyors who utilise sub-sea Remotely Operated Vehicles (ROVs), launched from a surface vessel and piloted over the pipeline. ROVs capture data from various sensors/instruments which are subsequently reviewed and interpreted by human operators, creating a log of event annotations; a slow, labour-intensive and costly process. The paper presents an automatic image annotation framework that identifies/classifies key events of interest in the video footage viz. exposure, burial, field joints, anodes and free spans. The reported methodology utilises transfer learning with a Deep Convolutional Neural Network (ResNet-50), fine-tuned on real-life, representative data from challenging sub-sea environments with low lighting conditions, sand agitation, sea-life and vegetation. The network outputs are configured to perform multi-label image classifications for the critical events. The annotation performance varies between 95.1{\%} and 13 99.7{\%} in terms of accuracy and 90.4{\%} and 99.4{\%} in terms of F1-Score depending on event type. The performance results are on a per-frame basis and corroborate the potential of the algorithm to be the foundation for an intelligent decision support framework that automates the annotation process. The solution can execute annotations in real-time and is significantly more cost-effective than human-only approaches.",
keywords = "visual inspection, subsea pipeline survey, multi-label image classification, deep learning, transfer learning",
author = "Anastasios Stamoulakatos and Javier Cardona and Chris McCaig and David Murray and Hein Filius and Robert Atkinson and Xavier Bellekens and Craig Michie and Ivan Andonovic and Pavlos Lazaridis and Andrew Hamilton and Hossain, {Md. Moinul} and {Di Caterina}, Gaetano and Christos Tachtatzis",
year = "2020",
month = "1",
day = "26",
doi = "10.3390/s20030674",
language = "English",
volume = "20",
journal = "Sensors",
issn = "1424-8220",
number = "3",

}

TY - JOUR

T1 - Automatic annotation of subsea pipelines using deep learning

AU - Stamoulakatos, Anastasios

AU - Cardona, Javier

AU - McCaig, Chris

AU - Murray, David

AU - Filius, Hein

AU - Atkinson, Robert

AU - Bellekens, Xavier

AU - Michie, Craig

AU - Andonovic, Ivan

AU - Lazaridis, Pavlos

AU - Hamilton, Andrew

AU - Hossain, Md. Moinul

AU - Di Caterina, Gaetano

AU - Tachtatzis, Christos

PY - 2020/1/26

Y1 - 2020/1/26

N2 - Regulatory requirements for sub-sea oil and gas operators mandates the frequent inspection of pipeline assets to ensure that their degradation and damage are maintained at acceptable levels. The inspection process is usually sub-contracted to surveyors who utilise sub-sea Remotely Operated Vehicles (ROVs), launched from a surface vessel and piloted over the pipeline. ROVs capture data from various sensors/instruments which are subsequently reviewed and interpreted by human operators, creating a log of event annotations; a slow, labour-intensive and costly process. The paper presents an automatic image annotation framework that identifies/classifies key events of interest in the video footage viz. exposure, burial, field joints, anodes and free spans. The reported methodology utilises transfer learning with a Deep Convolutional Neural Network (ResNet-50), fine-tuned on real-life, representative data from challenging sub-sea environments with low lighting conditions, sand agitation, sea-life and vegetation. The network outputs are configured to perform multi-label image classifications for the critical events. The annotation performance varies between 95.1% and 13 99.7% in terms of accuracy and 90.4% and 99.4% in terms of F1-Score depending on event type. The performance results are on a per-frame basis and corroborate the potential of the algorithm to be the foundation for an intelligent decision support framework that automates the annotation process. The solution can execute annotations in real-time and is significantly more cost-effective than human-only approaches.

AB - Regulatory requirements for sub-sea oil and gas operators mandates the frequent inspection of pipeline assets to ensure that their degradation and damage are maintained at acceptable levels. The inspection process is usually sub-contracted to surveyors who utilise sub-sea Remotely Operated Vehicles (ROVs), launched from a surface vessel and piloted over the pipeline. ROVs capture data from various sensors/instruments which are subsequently reviewed and interpreted by human operators, creating a log of event annotations; a slow, labour-intensive and costly process. The paper presents an automatic image annotation framework that identifies/classifies key events of interest in the video footage viz. exposure, burial, field joints, anodes and free spans. The reported methodology utilises transfer learning with a Deep Convolutional Neural Network (ResNet-50), fine-tuned on real-life, representative data from challenging sub-sea environments with low lighting conditions, sand agitation, sea-life and vegetation. The network outputs are configured to perform multi-label image classifications for the critical events. The annotation performance varies between 95.1% and 13 99.7% in terms of accuracy and 90.4% and 99.4% in terms of F1-Score depending on event type. The performance results are on a per-frame basis and corroborate the potential of the algorithm to be the foundation for an intelligent decision support framework that automates the annotation process. The solution can execute annotations in real-time and is significantly more cost-effective than human-only approaches.

KW - visual inspection

KW - subsea pipeline survey

KW - multi-label image classification

KW - deep learning

KW - transfer learning

U2 - 10.3390/s20030674

DO - 10.3390/s20030674

M3 - Article

VL - 20

JO - Sensors

JF - Sensors

SN - 1424-8220

IS - 3

M1 - 674

ER -