Dynamic-static unsupervised sequentiality, statistical subunits and lexicon for sign language recognition

Stavros Theodorakis, Vassilis Pitsikalis, Petros Maragos

Abstract: Highlights•A phonetic modeling approach for unsupervised sequentiality of dynamic-static SUs.•Model based sign segmentation and subunit modeling with HMMs.•Construction of a sign-to-subunits data-driven lexicon.•Comparison of different signers' pronunciations and unseen signer pronunciation compensation.•Evaluation on data from three corpora, two sign languages and unseen signers.