This project describes a computer based system that utilizes a 3D sensor to bridge the communication barrier between hearing (and/or speech) impaired people and hearing ones. Above figure shows how I love Vietnam demonstrated in Vietnamese Sign Language.
The proposed Vietnamese sign language reader successfully recognizes 28 static and 7 dynamic gestures taken from the Vietnam sign language dictionary. To recognize the gestures of a static type, various techniques have been deployed, such as Gabor Filtering, Fisher’s Discriminant Analysis and Cosine Metric Distance method.
The proposed technique achieves a good result with 93.89% accuracy and speed of 14 frames per second. Recognition of the dynamic gestures is based on the $1 Recognizer algorithm providing quite a good accuracy with 97.14% of accurate recognition with real-time running at 15 frames per second.
Demo
This is a quick demo video recorded on my laptop in Fall 2013, Australia.
Results
- This project is part of my thesis which can be read from HERE for further details.
- This project also gained a good feedbacks from my professors and my paper proposal finally got published in IEEE; if you want to read, the paper is assessable from RESEARCH GATE
- I published the source here for anyone interested in Vietnamese Sign Language reader with new technology.
https://github.com/vanducng/vietnamese-language-reader
Comments