Real-Time Visual Communication System Using Deep Residual Networks (ResNet50)

Effective communication remains a major challenge for individuals with hearing and speech impairments, especially when interacting with non-sign language users. To address this issue, intelligent systems capable of real-time translation are essential. This work presents a Smart Sign Language Translator designed to support seamless two-way communication using advanced deep learning techniques. The system employs ResNet50, a deep Convolutional Neural Network (CNN), to accurately identify hand gestures captured through a camera. The recognized gestures are translated into text and speech outputs, while spoken input is converted into corresponding sign language representations. The integration of residual learning in ResNet50 enhances feature extraction and improves recognition accuracy even in complex environments. The proposed system ensures efficient real-time processing, user-friendly interaction, and reliable performance. By combining gesture recognition, speech processing, and automated translation, the system enhances accessibility and supports inclusive communication for individuals with hearing and speech disabilities.