Sign Language Recognition using Linguistically Derived Sub-Units (bibtex)
by Helen Cooper, Richard Bowden
Abstract:
This work proposes to learn linguistically-derived sub-unit classifiers for sign language. The responses of these classifiers can be combined by Markov models, producing efficient sign-level recognition. Tracking is used to create vectors of hand positions per frame as inputs for sub-unit classifiers learnt using AdaBoost. Grid-like classifiers are built around specific elements of the tracking vector to model the placement of the hands. Comparative classifiers encode the positional relationship between the hands. Finally, binary-pattern classifiers are applied over the tracking vectors of multiple frames to describe the motion of the hands. Results for the sub-unit classifiers in isolation are presented, reaching averages over 90\%. Using a simple Markov model to combine the sub-unit classifiers allows sign level classifiation giving an average of 63\%, over a 164 sign lexicon, with no grammatical constraints.
Reference:
Helen Cooper, Richard Bowden, "Sign Language Recognition using Linguistically Derived Sub-Units", In Proceedings of the Language Resources and Evaluation Conference Workshop on the Representation and Processing of Sign Languages : Corpora and Sign Languages Technologies, Valetta, Malta, 2010. (poster(pdf))
Bibtex Entry:
@INPROCEEDINGS{Cooper_Sign_2010a,
  author = {Helen Cooper and Richard Bowden},
  title = {Sign Language Recognition using Linguistically Derived Sub-Units},
  booktitle = { Proceedings of the Language Resources and Evaluation Conference
	Workshop on the Representation and Processing of Sign Languages :
	Corpora and Sign Languages Technologies},
  year = {2010},
  address = {Valetta, Malta},
  month = may # {17 -- 23},
  abstract = {This work proposes to learn linguistically-derived sub-unit classifiers
	for sign language. The responses of these classifiers can be combined
	by Markov models, producing efficient sign-level recognition. Tracking
	is used to create vectors of hand positions per frame as inputs for
	sub-unit classifiers learnt using AdaBoost. Grid-like classifiers
	are built around specific elements of the tracking vector to model
	the placement of the hands. Comparative classifiers encode the positional
	relationship between the hands. Finally, binary-pattern classifiers
	are applied over the tracking vectors of multiple frames to describe
	the motion of the hands. Results for the sub-unit classifiers in
	isolation are presented, reaching averages over 90\%. Using a simple
	Markov model to combine the sub-unit classifiers allows sign level
	classifiation giving an average of 63\%, over a 164 sign lexicon,
	with no grammatical constraints.},
  comment = {<a target="_blank" href="papers/SLR-LinguisticallyDerivedSubUnits_poster.pdf">poster(pdf)</a>},
  doi = {10.1.1.168.6462},
  url = {http://personal.ee.surrey.ac.uk/Personal/H.Cooper/research/papers/SLR-LinguisticallyDerivedSubUnits.pdf}
}
Powered by bibtexbrowser