Sign to Text: Automated Sign Language Interpretation using LSTM and Computer Vision

Sign language serves as a crucial communication tool for individuals with speech or hearing impairments. Recent advancements in computer vision and deep learning offer opportunities to develop efficient, real-time sign language recognition systems that bridge communication gaps. This study proposes a sign language recognition model utilizing the Mediapipe model and Long Short-Term Memory (LSTM) networks to detect and interpret American Sign Language (ASL) gestures in real time. The proposed system extracts and analyzes key landmarks across the body, hands, and face, leveraging CNN-LSTM architecture to capture both spatial and temporal features essential for accurate gesture recognition. Evaluated on a diverse ASL dataset, the proposed model achieved an accuracy of 92%, demonstrating robustness and adaptability in varied environments. The proposed approach holds promise for applications in assistive technologies and educational tools, providing an accessible solution for sign language interpretation.