A Meta-Analytic Quantitative Review of Methods and Techniques for Sign Language Recognition Approaches

Abstract

While numerous machine learning methods and techniques have been employed for sign language recognition systems, there has not yet been a study that consolidates the methods to provide insight into trends and advancements in this domain. This gap remains as current reviews and survey papers do not adopt the format of a meta-analysis. Hence, the primary objective of this study was to fill this knowledge gap by conducting a meta-analysis and review of sign language recognition, deep learning, machine learning, and hybrid-based methods. The extraction of relevant articles was carried out following the technique of preferred reporting items for systematic reviews (PRISMA). Statistical analysis was performed using Stata version 15. The results of the meta-analysis showed an overall pooled estimate of 40%, with a range of 1% to 87% within a 95% confidence interval. The high pooled effect size of <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathrm{I}^{2}=95.78\%$</tex> suggests a significant statistical heterogeneity across all the studies.