Mean-shift and sparse sampling based SMC-PHD filtering for audio informed visual speaker tracking
- Submitting institution
-
Kingston University
- Unit of assessment
- 11 - Computer Science and Informatics
- Output identifier
- 11-14-1348
- Type
- D - Journal article
- DOI
-
10.1109/TMM.2016.2599150
- Title of journal
- IEEE Transactions on Multimedia
- Article number
- -
- First page
- 2417
- Volume
- 18
- Issue
- -
- ISSN
- 1520-9210
- Open access status
- Compliant
- Month of publication
- -
- 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
-
-
- Research group(s)
-
-
- Citation count
- 13
- Proposed double-weighted
- No
- Reserve for an output with double weighting
- No
- Additional information
- One significant drawback of filter based tracking algorithms is their heavy computational cost as large numbers of particles need to be distributed in each frame to model the state of the speakers. A significant advance reported by this paper is the use of audio information to reduce the number particles projected at each time step. Moreover, the computational complexity is further controlled with a sparse sampling technique. With increasing use of tracking algorithms on smaller mobile devices, the ability of reducing the computational complexity of those algorithms has greatly enhanced their suitability for real-life applications.
- Author contribution statement
- -
- Non-English
- No
- English abstract
- -