Nth layer Hierarchical Bidirectional LSTM sign language Interpretation for Hearing Impaired person

Multidimensional investigation into sign language interpreting systems improves recognition effectiveness while tackling the complexities and difficulties that the deaf and mute community faces. However, current models failed to validate accurate signed gestures and lacked expedited prediction analysis. The integration of Bidirectional LSTM models, while leveraging extensive temporal data, often resulted in the amplification of compounding errors in spatiotemporal analysis. To address this issue, we propose an N th layer Bidirectional LSTM (NHBLSTM) to solve compounding errors within a multidimensional sign language interpretation. NHBLSTM employs a co-attention prediction mechanism that refines sequential input at multiple levels, allowing for precise evaluation of dynamic data flow and minimizing errors. The model’s hierarchical bidirectional structure within the N th layer reduces training epochs, while its multi-tiered design enhances the validation of diverse gestures, leading to faster and more accurate predictions. Through experimental comparison with LSTM models, the NHBLSTM approach demonstrated a significant improvement, achieving 99% accuracy, and offering a robust solution for sign language interpretation, particularly benefiting individuals with hearing impairments.