Learning shape priors for object segmentation via neural networks

Simon Safar; Ming Hsuan Yang

doi:10.1109/ICIP.2015.7351118

Learning shape priors for object segmentation via neural networks

Simon Safar, Ming Hsuan Yang

College of Computing

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

8 Citations (Scopus)

Abstract

We present a joint algorithm for object segmentation that integrates both global shape and local edge information in a deep learning framework. The proposed architecture uses convolutional layers to extract image features, followed by a fully connected section to represent shapes specific to a given object class. This preliminary mask is further refined by matching segmentation mask patches to local features. These processing steps facilitate learning the shape priors effectively with a feedforward pass rather than complex inference methods. Furthermore, our novel convolutional refinement stage presents a convincing alternative to Conditional Random Fields, with promising results on multiple datasets.

Original language	English
Title of host publication	2015 IEEE International Conference on Image Processing, ICIP 2015 - Proceedings
Publisher	IEEE Computer Society
Pages	1835-1839
Number of pages	5
ISBN (Electronic)	9781479983391
DOIs	https://doi.org/10.1109/ICIP.2015.7351118
Publication status	Published - 2015 Dec 9
Event	IEEE International Conference on Image Processing, ICIP 2015 - Quebec City, Canada Duration: 2015 Sept 27 → 2015 Sept 30

Publication series

Name	Proceedings - International Conference on Image Processing, ICIP
Volume	2015-December
ISSN (Print)	1522-4880

Other

Other	IEEE International Conference on Image Processing, ICIP 2015
Country/Territory	Canada
City	Quebec City
Period	15/9/27 → 15/9/30

Bibliographical note

Publisher Copyright:
© 2015 IEEE.

All Science Journal Classification (ASJC) codes

Software
Computer Vision and Pattern Recognition
Signal Processing

Access to Document

10.1109/ICIP.2015.7351118

Cite this

@inproceedings{f73b89d425614a41bb062302ff1e3baa,

title = "Learning shape priors for object segmentation via neural networks",

abstract = "We present a joint algorithm for object segmentation that integrates both global shape and local edge information in a deep learning framework. The proposed architecture uses convolutional layers to extract image features, followed by a fully connected section to represent shapes specific to a given object class. This preliminary mask is further refined by matching segmentation mask patches to local features. These processing steps facilitate learning the shape priors effectively with a feedforward pass rather than complex inference methods. Furthermore, our novel convolutional refinement stage presents a convincing alternative to Conditional Random Fields, with promising results on multiple datasets.",

author = "Simon Safar and Yang, {Ming Hsuan}",

note = "Publisher Copyright: {\textcopyright} 2015 IEEE.; IEEE International Conference on Image Processing, ICIP 2015 ; Conference date: 27-09-2015 Through 30-09-2015",

year = "2015",

month = dec,

day = "9",

doi = "10.1109/ICIP.2015.7351118",

language = "English",

series = "Proceedings - International Conference on Image Processing, ICIP",

publisher = "IEEE Computer Society",

pages = "1835--1839",

booktitle = "2015 IEEE International Conference on Image Processing, ICIP 2015 - Proceedings",

address = "United States",

}

Safar, S & Yang, MH 2015, Learning shape priors for object segmentation via neural networks. in 2015 IEEE International Conference on Image Processing, ICIP 2015 - Proceedings., 7351118, Proceedings - International Conference on Image Processing, ICIP, vol. 2015-December, IEEE Computer Society, pp. 1835-1839, IEEE International Conference on Image Processing, ICIP 2015, Quebec City, Canada, 15/9/27. https://doi.org/10.1109/ICIP.2015.7351118

Learning shape priors for object segmentation via neural networks. / Safar, Simon; Yang, Ming Hsuan.
2015 IEEE International Conference on Image Processing, ICIP 2015 - Proceedings. IEEE Computer Society, 2015. p. 1835-1839 7351118 (Proceedings - International Conference on Image Processing, ICIP; Vol. 2015-December).

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

TY - GEN

T1 - Learning shape priors for object segmentation via neural networks

AU - Safar, Simon

AU - Yang, Ming Hsuan

PY - 2015/12/9

Y1 - 2015/12/9

N2 - We present a joint algorithm for object segmentation that integrates both global shape and local edge information in a deep learning framework. The proposed architecture uses convolutional layers to extract image features, followed by a fully connected section to represent shapes specific to a given object class. This preliminary mask is further refined by matching segmentation mask patches to local features. These processing steps facilitate learning the shape priors effectively with a feedforward pass rather than complex inference methods. Furthermore, our novel convolutional refinement stage presents a convincing alternative to Conditional Random Fields, with promising results on multiple datasets.

AB - We present a joint algorithm for object segmentation that integrates both global shape and local edge information in a deep learning framework. The proposed architecture uses convolutional layers to extract image features, followed by a fully connected section to represent shapes specific to a given object class. This preliminary mask is further refined by matching segmentation mask patches to local features. These processing steps facilitate learning the shape priors effectively with a feedforward pass rather than complex inference methods. Furthermore, our novel convolutional refinement stage presents a convincing alternative to Conditional Random Fields, with promising results on multiple datasets.

UR - http://www.scopus.com/inward/record.url?scp=84956688903&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84956688903&partnerID=8YFLogxK

U2 - 10.1109/ICIP.2015.7351118

DO - 10.1109/ICIP.2015.7351118

M3 - Conference contribution

AN - SCOPUS:84956688903

T3 - Proceedings - International Conference on Image Processing, ICIP

SP - 1835

EP - 1839

BT - 2015 IEEE International Conference on Image Processing, ICIP 2015 - Proceedings

PB - IEEE Computer Society

T2 - IEEE International Conference on Image Processing, ICIP 2015

Y2 - 27 September 2015 through 30 September 2015

ER -

Learning shape priors for object segmentation via neural networks

Abstract

Publication series

Other

Bibliographical note

All Science Journal Classification (ASJC) codes

Access to Document

Other files and links

Fingerprint

Cite this