Sign Language Recognition System For The Deaf Using Mediapipe Based On The Long-Short Term Memory Method

As the world becomes increasingly digital, it is essential to develop inclusive technologies that bridge communication gaps, especially for individuals with hearing impairments. This research proposes a sign language recognition system utilizing Mediapipe for hand detection and Long Short-Term Memory (LSTM) networks for gesture recognition. The primary goal of this study is to accurately track and classify hand movements associated with Indonesian Sign Language (BISINDO), focusing on two-handed gestures. By leveraging Mediapipe for real-time hand tracking and LSTM for recognizing temporal dynamics, this system is capable of recognizing sign language with high accuracy across varied environmental conditions. Experimental results demonstrate the system's robust performance, with testing accuracy reaching up to 95% under controlled conditions. This research contributes to the field of non-verbal communication by enhancing the accessibility of sign language recognition technology and provides a foundation for future advancements in pattern recognition for real-time applications. Additionally, it addresses the challenges of lighting, viewpoint variation, and environmental noise, offering potential solutions for the deployment of sign language recognition systems in real-world scenarios.