In The School of Computer Science & Software Engineering at The University of Western Australia, the Sign Language Translation Group has been actively working on two-way translation between Auslan (Australain Sign Language) and English.
This project: Automatic Auslan Finger-spelling Recognition using Hidden Markov Models was undertaken as part of Mr. Paul Goh's for his Honour project for the Bachelor of Computer and Mathematical Science. The Auslan Finger-spelling Recogniser (AFR) Sytem is capable of recognizing a vocabulary of 20 finger-spelling words and names in real-time, with an accuracy of 97.15% at the letter level and 88.61% at the word level. |
