Repository logo
  • Log In
    New user? Click here to register.Have you forgotten your password?
University College Dublin
    Colleges & Schools
    Statistics
    All of DSpace
  • Log In
    New user? Click here to register.Have you forgotten your password?
  1. Home
  2. College of Engineering & Architecture
  3. School of Mechanical and Materials Engineering
  4. Mechanical & Materials Engineering Research Collection
  5. Semantic Segmentation of Underwater Imagery Using Deep Networks Trained on Synthetic Imagery
 
  • Details
Options

Semantic Segmentation of Underwater Imagery Using Deep Networks Trained on Synthetic Imagery

Author(s)
O'Byrne, Michael  
Pakrashi, Vikram  
Schoefs, Franck  
Ghosh, Bidisha  
Uri
http://hdl.handle.net/10197/10346
Date Issued
2018-08-04
Date Available
2019-05-08T11:55:17Z
Abstract
Recent breakthroughs in the computer vision community have led to the emergence of efficient deep learning techniques for end-to-end segmentation of natural scenes. Underwater imaging stands to gain from these advances, however, deep learning methods require large annotated datasets for model training and these are typically unavailable for underwater imaging applications. This paper proposes the use of photorealistic synthetic imagery for training deep models that can be applied to interpret real-world underwater imagery. To demonstrate this concept, we look at the specific problem of biofouling detection on marine structures. A contemporary deep encoder–decoder network, termed SegNet, is trained using 2500 annotated synthetic images of size 960 × 540 pixels. The images were rendered in a virtual underwater environment under a wide variety of conditions and feature biofouling of various size, shape, and colour. Each rendered image has a corresponding ground truth per-pixel label map. Once trained on the synthetic imagery, SegNet is applied to segment new real-world images. The initial segmentation is refined using an iterative support vector machine (SVM) based post-processing algorithm. The proposed approach achieves a mean Intersection over Union (IoU) of 87% and a mean accuracy of 94% when tested on 32 frames extracted from two distinct real-world subsea inspection videos. Inference takes several seconds for a typical image.
Sponsorship
Science Foundation Ireland
Other Sponsorship
CAPACITES/IXEAD Society
Type of Material
Journal Article
Publisher
MDPI AG
Journal
Journal of Marine Science and Engineering
Volume
6
Issue
3
Start Page
1
End Page
15
Copyright (Published Version)
2018 the Authors
Subjects

Semantic segmentation...

Biofouling

Image processing

Underwater imaging

DOI
10.3390/jmse6030093
Language
English
Status of Item
Peer reviewed
ISSN
2077-1312
This item is made available under a Creative Commons License
https://creativecommons.org/licenses/by-nc-nd/3.0/ie/
File(s)
Loading...
Thumbnail Image
Name

jmse-331333-update.pdf

Size

4.48 MB

Format

Adobe PDF

Checksum (MD5)

9ed1a43002080ff605803572091f19a6

Owning collection
Mechanical & Materials Engineering Research Collection

Item descriptive metadata is released under a CC-0 (public domain) license: https://creativecommons.org/public-domain/cc0/.
All other content is subject to copyright.

Built with DSpace-CRIS software - Extension maintained and optimized by 4Science

  • Cookie settings
  • Privacy policy
  • End User Agreement