Exploring Sensor Fusion Techniques for Enhanced Dynamic Hand Gesture Recognition: A Comprehensive Metadata Analysis

Due to the widespread prevalence of hearing loss, it is crucial to develop reliable sign language recognition (SLR) technologies on a regional scale. Upon conducting an investigation into the most recent developments in machine learning and deep learning for automatically identifying sign languages, we realized that conceptual classification is essential for existing methods to correctly interpret the available data. Consequently, we focused on the features utilized by the majority of existing SLR techniques. We delineate the benefits and drawbacks of each approach and propose a research strategy for future studies in this article. Our research suggests that multimodal recognition, which integrates both visual and sensory channels, is preferable to traditional unimodal analysis. Moreover, recent advancements enable scientists to interpret continuous sign language with minimal latency, a significant improvement from merely recognizing individual symbols and sentences. While several models provided in the literature perform adequately across various tasks, none of them are yet generalizable enough for widespread commercial use. Nevertheless, the pace of research is promising, and further advancements are expected as certain challenges are addressed. In addition, it is important to note that future research in SLR should consider extending the scope to encompass regional variations in sign languages, as well as addressing the challenges associated with real-world applications, such as varying lighting conditions and background noise. By addressing these challenges and leveraging the latest advancements in machine learning and deep learning, we can continue to improve the accuracy and usability of SLR technologies, thereby enhancing accessibility for individuals with hearing impairments.