dc.contributor.author | De Ath, G | |
dc.contributor.author | Everson, R | |
dc.date.accessioned | 2019-05-21T09:59:36Z | |
dc.date.issued | 2018-12-13 | |
dc.description.abstract | Model initialisation is an important component of object tracking. Tracking algorithms are generally provided with the first frame of a sequence and a bounding box (BB) indicating the location of the object. This BB may contain a large number of background pixels in addition to the object and can lead to parts-based tracking algorithms initialising their object models in background regions of the BB. In this paper, we tackle this as a missing labels problem, marking pixels sufficiently away from the BB as belonging to the background and learning the labels of the unknown pixels. Three techniques, One-Class SVM (OC-SVM), Sampled-Based Background Model (SBBM) (a novel background model based on pixel samples), and Learning Based Digital Matting (LBDM), are adapted to the problem. These are evaluated with leave-one-video-out cross-validation on the VOT2016 tracking benchmark. Our evaluation shows both OC-SVMs and SBBM are capable of providing a good level of segmentation accuracy but are too parameter-dependent to be used in real-world scenarios. We show that LBDM achieves significantly increased performance with parameters selected by cross validation and we show that it is robust to parameter variation. | en_GB |
dc.identifier.citation | 2018 15th Conference on Computer and Robot Vision (CRV), 8-10 May 2018, Toronto, Canada, pp. 142 - 149 | en_GB |
dc.identifier.doi | 10.1109/CRV.2018.00029 | |
dc.identifier.uri | http://hdl.handle.net/10871/37166 | |
dc.language.iso | en | en_GB |
dc.publisher | Institute of Electrical and Electronics Engineers (IEEE) | en_GB |
dc.relation.url | http://github.com/georgedeath/initialisation-problem | en_GB |
dc.rights | © 2018 IEEE | en_GB |
dc.subject | Feature extraction | en_GB |
dc.subject | Support vector machines | en_GB |
dc.subject | Image segmentation | en_GB |
dc.subject | Image color analysis | en_GB |
dc.subject | Adaptation models | en_GB |
dc.subject | Kernel | en_GB |
dc.subject | Visualization | en_GB |
dc.title | Visual object tracking: The initialisation problem | en_GB |
dc.type | Article | en_GB |
dc.date.available | 2019-05-21T09:59:36Z | |
dc.description | This is the author accepted manuscript. The final version is available from IEEE via the DOI in this record | en_GB |
dc.description | Source code for all three methods is publicly available at:
http://github.com/georgedeath/initialisation-problem | en_GB |
dc.rights.uri | http://www.rioxx.net/licenses/all-rights-reserved | en_GB |
dcterms.dateAccepted | 2018-03-11 | |
rioxxterms.version | AM | en_GB |
rioxxterms.licenseref.startdate | 2018-12-13 | |
rioxxterms.type | Conference Paper/Proceeding/Abstract | en_GB |
refterms.dateFCD | 2019-05-21T09:39:36Z | |
refterms.versionFCD | AM | |
refterms.dateFOA | 2019-05-21T09:59:39Z | |
refterms.panel | B | en_GB |