A survey on sign language recognition from perspectives of traditional and deep-learning methods
Tangfei Tao,Yizhe Zhao,Jieli Zhu,Tianyu Liu,Jiachen Kuang
DOI: https://doi.org/10.1016/j.jvcir.2024.104363
IF: 2.887
2024-12-08
Journal of Visual Communication and Image Representation
Abstract:The deaf, who lose the ability to learn spoken language due to hearing impairment at a young age, is a large social group with significant influence in society. Unfortunately, their sign language can only be used within the deaf community as others do not understand it. Thus, the sign language recognition (SLR) technique comes into being to help normal people understand the meaning of sign language. Although SLR is of great importance, many of the existing reviews fail to give a comprehensive introduction from perspectives of methods and datasets in this field. In recent years, new methods of sign language recognition have been developed and achieved with good results, so it is necessary to make a summarization. This review mainly focuses on the introduction of sign language recognition techniques based on algorithms especially in recent years, including the sign language datasets, recognition models based on traditional methods and deep learning approaches, and existing problems in SLR. The basic principles of the various methods are presented and the characteristics of the methods and datasets are also compared. We hope that this review will provide some reference and help for future research on sign language recognition.
computer science, information systems, software engineering