Advanced

Max-margin learning of deep structured models for semantic segmentation

Larsson, Måns; Alvén, Jennifer and Kahl, Fredrik LU (2017) 20th Scandinavian Conference on Image Analysis, SCIA 2017 In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) 10270 LNCS. p.28-40
Abstract

During the last few years most work done on the task of image segmentation has been focused on deep learning and Convolutional Neural Networks (CNNs) in particular. CNNs are powerful for modeling complex connections between input and output data but lack the ability to directly model dependent output structures, for instance, enforcing properties such as smoothness and coherence. This drawback motivates the use of Conditional Random Fields (CRFs), widely applied as a post-processing step in semantic segmentation. In this paper, we propose a learning framework that jointly trains the parameters of a CNN paired with a CRF. For this, we develop theoretical tools making it possible to optimize a max-margin objective with back-propagation.... (More)

During the last few years most work done on the task of image segmentation has been focused on deep learning and Convolutional Neural Networks (CNNs) in particular. CNNs are powerful for modeling complex connections between input and output data but lack the ability to directly model dependent output structures, for instance, enforcing properties such as smoothness and coherence. This drawback motivates the use of Conditional Random Fields (CRFs), widely applied as a post-processing step in semantic segmentation. In this paper, we propose a learning framework that jointly trains the parameters of a CNN paired with a CRF. For this, we develop theoretical tools making it possible to optimize a max-margin objective with back-propagation. The max-margin loss function gives the model good generalization capabilities. Thus, the method is especially suitable for applications where labelled data is limited, for example, medical applications. This generalization capability is reflected in our results where we are able to show good performance on two relatively small medical datasets. The method is also evaluated on a public benchmark (frequently used for semantic segmentation) yielding results competitive to state-of-the-art. Overall, we demonstrate that end-to-end max-margin training is preferred over piecewise training when combining a CNN with a CRF.

(Less)
Please use this url to cite or link to this publication:
author
organization
publishing date
type
Chapter in Book/Report/Conference proceeding
publication status
published
subject
keywords
Convolutional Neural Networks, Markov random fields, Segmentation
in
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
volume
10270 LNCS
pages
13 pages
publisher
Springer Verlag
conference name
20th Scandinavian Conference on Image Analysis, SCIA 2017
external identifiers
  • scopus:85020436836
ISSN
16113349
03029743
ISBN
9783319591285
DOI
10.1007/978-3-319-59129-2_3
language
English
LU publication?
yes
id
b59f0d9c-c138-4c55-851a-c17815f958b6
date added to LUP
2017-06-30 09:01:19
date last changed
2017-06-30 09:01:19
@inproceedings{b59f0d9c-c138-4c55-851a-c17815f958b6,
  abstract     = {<p>During the last few years most work done on the task of image segmentation has been focused on deep learning and Convolutional Neural Networks (CNNs) in particular. CNNs are powerful for modeling complex connections between input and output data but lack the ability to directly model dependent output structures, for instance, enforcing properties such as smoothness and coherence. This drawback motivates the use of Conditional Random Fields (CRFs), widely applied as a post-processing step in semantic segmentation. In this paper, we propose a learning framework that jointly trains the parameters of a CNN paired with a CRF. For this, we develop theoretical tools making it possible to optimize a max-margin objective with back-propagation. The max-margin loss function gives the model good generalization capabilities. Thus, the method is especially suitable for applications where labelled data is limited, for example, medical applications. This generalization capability is reflected in our results where we are able to show good performance on two relatively small medical datasets. The method is also evaluated on a public benchmark (frequently used for semantic segmentation) yielding results competitive to state-of-the-art. Overall, we demonstrate that end-to-end max-margin training is preferred over piecewise training when combining a CNN with a CRF.</p>},
  author       = {Larsson, Måns and Alvén, Jennifer and Kahl, Fredrik},
  booktitle    = {Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)},
  isbn         = {9783319591285},
  issn         = {16113349},
  keyword      = {Convolutional Neural Networks,Markov random fields,Segmentation},
  language     = {eng},
  pages        = {28--40},
  publisher    = {Springer Verlag},
  title        = {Max-margin learning of deep structured models for semantic segmentation},
  url          = {http://dx.doi.org/10.1007/978-3-319-59129-2_3},
  volume       = {10270 LNCS},
  year         = {2017},
}