This review paper presents a comprehensive study of deep learning frameworks for real-time sign language detection. The paper examines major architectural approaches including Convolutional Neural Networks (CNNs), Recurrent Neural Networks (RNNs/LSTMs), Transformer-based models, Graph Neural Networks, and skeleton-based recognition systems. The review discusses their strengths, limitations, deployment challenges, security considerations, privacy concerns, dataset design methodologies, evaluation metrics, and real-world applications in healthcare, education, accessibility, and smart environments. Furthermore, the paper highlights emerging challenges such as adversarial robustness, data privacy, signer-independent evaluation, and ethical AI governance. The study aims to provide researchers and practitioners with a practical understanding of current sign language detection technologies and future research directions toward reliable and inclusive communication systems.
Sonekar et al. (Sat,) studied this question.