Name
Abstract | ||
|---|---|---|
As we are aware, there are millions of deaf-mutes around the world. It is a necessity to conduct research into sign language recognition as it is of massive significance to helping normal people and deaf-mute people communicate smoothly with others. A behavior recognition method was proposed in this paper to address the sign language recognition issue. Inspired by Multi-Fiber Networks, CBAM-ResNet neural network that extended the network structure of ResNet into 3D convolution was proposed and convolutional block attention module was added. In the fifth layer of the network, the structure unit of 3D-Res2Net was used to preserve the advantages possessed by the network structure of Multi-Fiber Networks, which could compensate for the deficiency of multi-Fiber Networks channel information fusion. A comparison was drawn of it with the models adding convolutional block attention module, Convolution Long Short-Term Memory, optical flow and other methods at the same time. The proposed method achieved an accuracy rate of 83.3% without optical flow on Chinese Sign Language Recognition Dataset, which was about 9% higher than that of Multi-Fiber Networks.
|
Year | DOI | Venue |
|---|---|---|
2019 | 10.1145/3358331.3358379 | Proceedings of the 2019 International Conference on Artificial Intelligence and Advanced Manufacturing |
Keywords | Field | DocType |
3D-Res2Net, Neural Networks, ResNet, Sign Language Recognition, convolutional block attention module | Computer science,Speech recognition,Sign language,Residual neural network | Conference |
ISBN | Citations | PageRank |
978-1-4503-7202-2 | 0 | 0.34 |
References | Authors | |
0 | 3 | |
Name | Order | Citations | PageRank |
|---|---|---|---|
| Huang Chao | 1 | 0 | 0.34 |
| Wang Fenhua | 2 | 0 | 0.34 |
| Ran Zhang | 3 | 33 | 13.46 |