Tamil sign language using relational bilevel aggregation graph convolutional network

Automatic translation of sign language to text facilitates communication between deaf or mute persons and others, including those who are not comfortable with sign language. In this manuscript, Identification of Tamil sign language utilising relational bilevel aggregation graph convolutional network (IDN-TSL-RBAGCN) is proposed. Initially, the data is collected through Tamil sign language gesture images. Artificial lizard search optimisation algorithm (ALSOA) is employed to enhance the weight parameter of relational bilevel aggregation graph convolutional network classifier (RBAGCN), which precisely classifies the Tamil sign language from the identified pattern. The proposed IDN-TSL-RBAGCN method attains 28.76%, 33.68% and 21.78% higher accuracy when compared with existing methods, like Indian sign language recognition utilising wearable sensors with multiple label categorisation (ISL-MLC), deep learning-dependent sign language recognition system for static signs (SLR-DL), and real-time vernacular sign language recognition utilising media pipe and machine learning (SLR-ML) respectively.