AI-Powered Sign Language Translation: Enhancing Communication with MediaPipe and Machine Learning

Abstract

The rapid advancement of artificial intelligence (AI) and machine learning has transformed communication accessibility for deaf and hard-of-hearing individuals. This research investigates using Media Pipe and Long Short-Term Memory (LSTM) networks for real-time sign language translation, focusing on overcoming barriers in Arabic Sign Language (ArSL). The study employs a landmark-driven approach, extracting hand and pose features, achieving 99.46% accuracy on a 10,000-sample dataset. Despite its success, challenges include handling diverse signing styles and environmental variations, with limitations in low-light conditions and subtle gesture recognition. TensorFlow Lite enables efficient mobile deployment, though technical details like data storage formats are omitted for brevity. Future work will explore dataset expansion, enhanced temporal modeling, and multimodal integration to boost accuracy and accessibility. Ethical considerations underscore the need for responsible AI development in assistive technologies.