<![CDATA[Sign language plays a crucial role in facilitating communication among individuals with hearing impairments. In Indonesia, the deaf community often rely on BISINDO (Indonesian Sign Language) to communicate amongst themselves. People who are unfamiliar with sign language will face difficulties. This research aims to develop a system for recognizing sign language using geometric features extracted from hand joint coordinates using Google's MediaPipe Hands framework. The dataset contains 12 common words, each recorded 30 times with 30 frames recorded for each instance. This will facilitate communication between deaf and hearing individuals. We conducted tests on LSTM- Geometric and CNN1D- Geometric models using geometric features, and on CNN-LSTM-Spatial and CNN1D-LSTM-Spatial models using spatial features. The results indicate that the LSTM model with geometric features achieved the highest accuracy of 99%. Geometric features have been shown to be more effective than spatial features for classifying sign language gestures.]]>
Copyrights © 2024
