Enriching texture analysis with semantic data
Enriching texture analysis with semantic data
We argue for the importance of explicit semantic modelling in human-centred texture analysis tasks such as retrieval, annotation, synthesis, and zero-shot learning.
To this end, low-level attributes are selected and used to define a semantic space for texture.
319 texture classes varying in illumination and rotation are positioned within this semantic space using a pairwise relative comparison procedure. Visual features used by existing texture descriptors are then assessed in terms of their correspondence to the semantic space. Textures with strong presence of attributes connoting randomness and complexity are shown to be poorly modelled by existing descriptors.
In a retrieval experiment semantic descriptors are shown to outperform visual descriptors. Semantic modelling of texture is thus shown to provide considerable value in both feature selection and in analysis tasks.
computer vision, texture, semantics
Matthews, Tim
f41aa009-4f12-4887-8427-50d344d5d9b3
Nixon, Mark S.
2b5b9804-5a81-462a-82e6-92ee5fa74e12
Niranjan, Mahesan
5cbaeea8-7288-4b55-a89c-c43d212ddd4f
25 June 2013
Matthews, Tim
f41aa009-4f12-4887-8427-50d344d5d9b3
Nixon, Mark S.
2b5b9804-5a81-462a-82e6-92ee5fa74e12
Niranjan, Mahesan
5cbaeea8-7288-4b55-a89c-c43d212ddd4f
Matthews, Tim, Nixon, Mark S. and Niranjan, Mahesan
(2013)
Enriching texture analysis with semantic data.
IEEE Conference on Computer Vision and Pattern Recognition, Portland, United States.
25 - 27 Jun 2013.
Record type:
Conference or Workshop Item
(Paper)
Abstract
We argue for the importance of explicit semantic modelling in human-centred texture analysis tasks such as retrieval, annotation, synthesis, and zero-shot learning.
To this end, low-level attributes are selected and used to define a semantic space for texture.
319 texture classes varying in illumination and rotation are positioned within this semantic space using a pairwise relative comparison procedure. Visual features used by existing texture descriptors are then assessed in terms of their correspondence to the semantic space. Textures with strong presence of attributes connoting randomness and complexity are shown to be poorly modelled by existing descriptors.
In a retrieval experiment semantic descriptors are shown to outperform visual descriptors. Semantic modelling of texture is thus shown to provide considerable value in both feature selection and in analysis tasks.
Text
main.pdf
- Version of Record
More information
Accepted/In Press date: 16 April 2013
Published date: 25 June 2013
Venue - Dates:
IEEE Conference on Computer Vision and Pattern Recognition, Portland, United States, 2013-06-25 - 2013-06-27
Keywords:
computer vision, texture, semantics
Organisations:
Vision, Learning and Control
Identifiers
Local EPrints ID: 352685
URI: http://eprints.soton.ac.uk/id/eprint/352685
PURE UUID: 09163fd9-4c60-4448-bc29-3742a18ecf92
Catalogue record
Date deposited: 20 May 2013 14:20
Last modified: 15 Mar 2024 03:29
Export record
Contributors
Author:
Tim Matthews
Author:
Mahesan Niranjan
Download statistics
Downloads from ePrints over the past year. Other digital versions may also be available to download e.g. from the publisher's website.
View more statistics