Video-Based Continuous Sign Language Recognition Using Statistical Methods

This paper is concerned with a development of a videobased recognition system of continuous sign language. The system aimsfor an automatic signer dependent recognition of sign language sentences, based on a lexicon of 97 signs of GermanSign Language (GSL).The recognition system is based on Hidden Markov Models with one model for each sign. A single video camera is utilisedfor data acquisition. Beamsearch is employedfor the recognition task. For a better result a language model is implemented, which is able to handle a-priori knowledge ofthe training corpus. Difserent results are givenfor a vocabulary of 52 respectively 97 signs with digerent employed language models (Unigram and Bigram). The system achieves an accuracy of 913% based on a lexicon of 97 signs without a language model and 93.2% with employed Bigrams.

• Type: Conference
• Year: 2000
• Where: ICPR
• Authors: Britta Bauer, Hermann Hienz, Karl-Friedrich Kraiss
• Keywords: Continuous Sign Language, ICPR 2000, Language Model, Sign Language