Sign Language Translator for SIBI

Deaf people face communication barriers and sign language is essential for them, especially in Indonesia. Indonesia has two main sign languages: BISINDO and SIBI. BISINDO is community developed, while SIBI is adapted from American Sign Language for standardization. In this study, we develop a gesture recognition system to translate SIBI using a convolutional neural network (CNN). Utilizing MobileNetV2, our model achieved over 99.7<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">%</sup> accuracy in controlled tests and over 99.9% accuracy on mobile devices. However, live camera tests revealed challenges with accuracy across different scripts. Further optimization is needed. This work advances SIBI sign language recognition and enhances communication tools for the deaf community in Indonesia.