Fast 2D/3D object representation with growing neural gas
- Submitting institution
-
The University of Westminster
- Unit of assessment
- 11 - Computer Science and Informatics
- Output identifier
- 9y32x
- Type
- D - Journal article
- DOI
-
10.1007/s00521-016-2579-y
- Title of journal
- Neural Computing and Applications
- Article number
- -
- First page
- 903
- Volume
- 29
- Issue
- 10
- ISSN
- 0941-0643
- Open access status
- Compliant
- Month of publication
- September
- Year of publication
- 2016
- URL
-
-
- Supplementary information
-
-
- Request cross-referral to
- -
- Output has been delayed by COVID-19
- No
- COVID-19 affected output statement
- -
- Forensic science
- No
- Criminology
- No
- Interdisciplinary
- No
- Number of additional authors
-
4
- Research group(s)
-
-
- Citation count
- 7
- Proposed double-weighted
- No
- Reserve for an output with double weighting
- No
- Additional information
- Establishing optimal topology mapping of rigid and non-rigid models without compromising quality, by dynamically learning features of linear and non-linear input spaces and their entities, is an essential step in the 2D/3D reconstruction and tracking of objects. The significance of this paper is the automatic definition of an optimal number of nodes, without overfitting or underfitting the network, by optimising a minimum description length criterion, thus tracking objects locally wherever common regions are found. Methods similar to this can be used in real-time systems to accelerate many high performance experiments on graphics cards.
- Author contribution statement
- -
- Non-English
- No
- English abstract
- -
