Neural network based scheme for unsupervised video object segmentation

Doulamis Anastasios, D; Doulamis Nikolaos, D; Kollias Stefanos, D

dc.contributor.author	Doulamis Anastasios, D	en
dc.contributor.author	Doulamis Nikolaos, D	en
dc.contributor.author	Kollias Stefanos, D	en
dc.date.accessioned	2014-03-01T02:41:30Z
dc.date.available	2014-03-01T02:41:30Z
dc.date.issued	1998	en
dc.identifier.uri	https://dspace.lib.ntua.gr/xmlui/handle/123456789/30504
dc.subject	Map Estimation	en
dc.subject	Procedural Modeling	en
dc.subject	Training Algorithm	en
dc.subject	Video Conferencing	en
dc.subject	Video Object Segmentation	en
dc.subject	Markov Random Field	en
dc.subject	Neural Network	en
dc.subject.other	Learning algorithms	en
dc.subject.other	Mathematical models	en
dc.subject.other	Maximum likelihood estimation	en
dc.subject.other	Neural networks	en
dc.subject.other	Object recognition	en
dc.subject.other	Random processes	en
dc.subject.other	Video conferencing	en
dc.subject.other	Video telephone equipment	en
dc.subject.other	Markov random field	en
dc.subject.other	Maximum a posteriori estimation	en
dc.subject.other	Video object segmentation	en
dc.subject.other	Image segmentation	en
dc.title	Neural network based scheme for unsupervised video object segmentation	en
heal.type	conferenceItem	en
heal.identifier.primary	10.1109/ICIP.1998.723557	en
heal.identifier.secondary	http://dx.doi.org/10.1109/ICIP.1998.723557	en
heal.publicationDate	1998	en
heal.abstract	In this paper, we proposed a neural network based scheme for performing unsupervised video object segmentation, especially for videophone or video-conferencing applications. The procedure includes (a) a training algorithm for adapting the network weights to the current condition, (b) a Maximum A Posteriori (MAP) estimation procedure for optimally selecting the most representative data of the current environment as retraining data and (c) a decision mechanism for determining when network retraining should be activated. The training algorithm takes into consideration both the former and the current network knowledge in order to achieve good generalization. The MAP estimation procedure models the network output as a Markov Random Field (MRF) and optimally selects the set of training inputs and corresponding desired outputs, using initial estimates of human face and body. Finally, a verification mechanism is introduced which augments the training data, exploiting information of the previous and current environment.	en
heal.publisher	IEEE Comp Soc, Los Alamitos, CA, United States	en
heal.journalName	IEEE International Conference on Image Processing	en
dc.identifier.doi	10.1109/ICIP.1998.723557	en
dc.identifier.volume	2	en
dc.identifier.spage	632	en
dc.identifier.epage	636	en