File(s) under permanent embargo
Reason: the publisher does not allow archiving of the accepted manuscript.
Chinese Sign Language Fingerspelling Recognition via Six-Layer Convolutional Neural Network with Leaky Rectified Linear Units for Therapy and Rehabilitation
journal contribution
posted on 2020-03-27, 12:29 authored by Xianwei Jiang, Yu-Dong ZhangWith hundreds of millions of hearing-impaired people around the world, sign language is considered as an effective solution to help bring the patients into the society. Besides, sign language is important in speech therapy and rehabilitation. However, it is not easy to learn. Automatic sign language recognition has become the focus of attention, which can help save labor resources in hospitals and certain occasions. Based on computer vision, we proposed a six-layer deep convolutional neural network with batch normalization, leaky ReLU and dropout techniques to identify Chinese sign language fingerspelling. Experiments showed that our approach achieved an overall accuracy of 88.10 ± 1.48%. The result is superior to three state-of-the-art approaches.