Name
Abstract | ||
|---|---|---|
In order to build a sign language recognition framework, one needs to collect sign databases that contain multiple samples of isolated signs, which is a hard and time consum- ing task. In this study, our aim is to obtain such a database by automatically extracting isolated signs from continuou s signing, recorded from the broadcast news for the hearing- impaired. We present an unsupervised, multiple alignment- based approach for sign segmentation. Among the modal- ities used to form a sign, hand gestures carry most of the information, manifested as hand motion and shape. To han- dle these two sources of information, we experimented with different feature sets, with different fusion methods on di f- ferent alignment approaches: feature concatenation on Dy- namic Time Warping (DTW) and Hidden Markov Models (HMMs), modeling via coupled and parallel HMMs, and sequential fusion of DTW and HMM. Our experiments on Turkish broadcast news videos show that (1) using low level shape descriptors is suitable for the alignment task, (2) th e highest accuracy is obtained by modeling the signs with HMM using the intervals found previously by DTW. |
Year | DOI | Venue |
|---|---|---|
2009 | 10.1109/ICCVW.2009.5457527 | international conference on computer vision |
Keywords | Field | DocType |
gesture recognition,handicapped aids,hidden Markov models,image segmentation,natural language processing,automatic sign segmentation,continuous signing,dynamic time warping,feature concatenation,hearing-impaired,hidden Markov models,multiple alignment-based approach,multiple sequence alignment,sequential fusion,sign databases,sign language recognition,unsupervised approach | Pattern recognition,Dynamic time warping,Computer science,Gesture,Segmentation,Gesture recognition,Image segmentation,Speech recognition,Feature extraction,Sign language,Artificial intelligence,Hidden Markov model | Conference |
Volume | Issue | ISBN |
2009 | 1 | 978-1-4244-4441-0 |
Citations | PageRank | References |
1 | 0.35 | 16 |
Authors | ||
4 | ||
Name | Order | Citations | PageRank |
|---|---|---|---|
| Pinar Santemiz | 1 | 3 | 1.39 |
| Oya Aran | 2 | 386 | 26.91 |
| Murat Saraclar | 3 | 669 | 62.91 |
| lale akarun | 4 | 1201 | 70.68 |