Real-Time American Sign Language Recognition from Video Using Hidden Markov Models

Thad Starner and Alex Pentland

earlier version in ISCV '95
Coral Gables, FL

Hidden Markov models (HMM's) have been used prominently and successfully in speech recognition and, more recently, in handwriting recognition. Consequently, they seem ideal for visual recognition of complex, structured hand gestures such as are found in sign language. We describe two experiments that demonstrate a real-time HMM-based system for recognizing sentence level American Sign Language (ASL) without explicitly modeling the fingers. The first experiment tracks hands wearing colored gloves and attains a word accuracy of 99%. The second experiment tracks hands without gloves and attains a word accuracy of 92%. Both experiments have a 40 word lexicon.