Technological solutions for sign language recognition: a scoping review of research trends, challenges, and opportunities

Abstract

Sign languages are critical in conveying meaning by the use of a visual-manual modality and are the primary means of communication of the deaf and hard of hearing with their family members and with the society. With the advances in computer graphics, computer vision, neural networks, and the introduction of new powerful hardware, the research into sign languages has shown a new potential. Novel technologies can help people learn, communicate, interpret, translate, visualize, document, and develop various sign languages and their related skills. This paper reviews the technological advancements applied in sign language recognition, visualization, and synthesis. We defined multiple research questions to identify the underlying technological drivers that strive to improve the challenges in this domain. This study is designed in accordance with the PRISMA methodology. We searched for articles published between 2010 and 2021 in multiple digital libraries (i.e., Elsevier, Springer, IEEE, PubMed, and MDPI). To automate the initial steps of PRISMA for identifying potentially relevant articles, duplicate removal and basic screening, we utilized a Natural Language Processing toolkit. Then, we performed a synthesis of the existing body of knowledge and identified the different studies that achieved significant advancements in sign language recognition, visualization, and synthesis. The identified trends based on analysis of almost 2000 papers clearly show that technology developments, especially in image processing and deep learning, are driving new applications and tools that improve the various performance metrics in these sign language-related task. Finally, we identified which techniques and devices contribute to such results and what are the common threads and gaps that would open new research directions in the field.