A Comparative Study of Continuous Sign Language Recognition Techniques

Some CSLR systems depend Sign language serves as a critical means of communication on the skeleton or pose information as an input to the following delivered using visual cues, including hand gestures and facial stages, therefore, they extract the pose information from each expressions [1]. Sign language recognition (SLR) involves video's frame at the preprocessing stage. However, this stage interpreting signs in video sequences and converting them into may be skipped by some CSLR systems that utilize sensorbased their corresponding glosses. The process includes capturing systems for signs capturing [5]. The spatial feature the movements of the signer's hands and body which are extractor captures feature representations from sign's frames usually integrated with facial expressions [2]. SLR aims to using spatial information learning techniques, such as 2D facilitate communication between deaf individuals and the Convolutional Neural Networks (CNN), 3D CNNs, Graph broader community by converting sign language into a form Convolutional Networks (GCN) or Vision Transformers (ViT).