Enhancing gesture-assisted sign language recognition technique for hearing and impaired people using computer vision and artificial intelligence techniques

Sign language (SL) serves as a vital communication tool for individuals with hearing impairments, enabling them to convey thoughts and emotions. However, a major challenge arises when communicating with people unfamiliar with SL, leading to misinterpretations and communication barriers. To address this, gesture recognition techniques integrated with deep learning (DL) are being explored to translate SL into meaningful sentences. This study proposes an advanced system named Enhancing Gesture Recognition Techniques for Hearing and Speaking Impaired People using Computer Vision and Artificial Intelligence Techniques (EGRTHSIP-CVAIT), designed to facilitate seamless communication between hearing-impaired individuals and the broader community. The system employs median filtering (MF) to eliminate noise while preserving image details, followed by feature extraction using the EfficientNetB7 model, known for its performance in handling complex visual data. For gesture recognition, a Deep Belief Network (DBN) classifier is utilized. To enhance accuracy and computational efficiency, the model parameters are optimized using a novel hybrid of White Shark Optimizer and Whale Optimization Algorithm (WSO-WOA). Simulation results demonstrate that EGRTHSIP-CVAIT outperforms existing methods in terms of precision and robustness. This innovative approach significantly improves gesture recognition capabilities, thereby promoting better understanding and interaction for individuals with hearing and speaking impairments.