Sign Language Recognition using Deep Learning: A Review of Methods and Future Directions

Computer Engineering

Sign language plays a vital role in facilitating communication for individuals who are hearing-impaired or have speech difficulties. The increasing integration of artificial intelligence and neural networks has greatly accelerated the progress of sign language recognition systems, effectively closing the communication gap between hearing and non-hearing individuals. This paper reviews three key approaches leveraging neural networks: static gesture recognition using convolutional neural networks (cnns), dynamic sequence modeling with long short-term memory (lstm) networks, and real-time gesture processing through mediapipe-integrated architectures. While convolutional neural networks excel at recognizing static gestures, long short-term memory networks demonstrate superior capabilities in handling sequential and dynamic gestures. Each approach is assessed based on its methodology, the data used, its accuracy, and how well it can be applied in real-life situations. The review points out important areas that require further investigation, such as the variety of datasets, the ability to withstand environmental changes, and the efficiency of computational processes, emphasizing the importance of combining different types of data and developing scalable systems. Looking ahead, the focus is on incorporating transformers, lightweight models for edge devices, and seamless integration into wearable and virtual platforms. By overcoming current constraints and embracing cutting-edge technologies, this field has the potential to significantly improve accessibility and promote inclusivity for individuals with hearing impairments.

This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.