Advances and Challenges in Gesture Recognition and Sign Language Recognition: A Comprehensive Review

Gesture and sign language recognition are critical components in enabling inclusive communication, particularly for the deaf and hard-of-hearing communities. However, current recognition systems still face significant limitations such as poor generalization to real-world environments, limited training data, and high computational demands. This study aims to provide a comprehensive Systematic Literature Review (SLR) to analyze recent developments, identify technical gaps, and propose future directions in gesture and sign language recognition using deep learning technologies. Utilizing PRISMA guidelines, we analyzed 45 selected studies published between 2021 and 2024, sourced from Scopus using defined search criteria and inclusion parameters. The review highlights the emergence of hybrid deep learning models (e.g., CNN-LSTM) for capturing spatial-temporal features, the use of GANs for data augmentation to overcome dataset scarcity, and YOLO-based models for real-time applications. We also identified performance gaps in handling lighting variability and user diversity. Unlike prior reviews, this study contributes by synthesizing methods across vision and sensor-based modalities and proposing a multimodal integration framework as a research opportunity. The results suggest that deep learning has significantly improved recognition accuracy (up to 99.75%), yet challenges in deployment and adaptability remain. This review provides actionable insights for researchers and developers in building more robust and accessible gesture recognition systems.