A Comprehensive Review of Machine Learning and Deep Learning Techniques for Sign Language Recognition.

Abstract

Sign language serves as the primary means of communication for the deaf and hard-of-hearing community. However, communication barriers still exist between sign language users and the general population. Recent advances in machine learning (ML) and deep learning (DL) have enabled automated sign language recognition (SLR) systems capable of translating gestures into text or speech. This paper presents a systematic review of ML- and DL-based techniques used for SLR across various languages, including American Sign Language and Arabic Sign Language. The review examines classical machine learning approaches, convolutional neural networks (CNNs), hybrid architectures such as CNN-LSTM and CNN-HMM, and video-based models including 3D CNNs. Comparative analysis indicates that deep learning methods significantly outperform traditional techniques due to their ability to automatically extract spatial and temporal features. However, challenges such as limited datasets, signer variability, computational complexity, and real-time deployment constraints remain significant. This review identifies research gaps and discusses future directions, including multimodal learning, lightweight architectures, and multilingual datasets, to improve the scalability and applicability of SLR systems.