<![CDATA[Deaf individuals in Indonesia face challenges in using voice-based technology. This study aims to develop an interactive kiosk system utilizing hand gesture detection based on Long Short-Term Memory (LSTM) to provide a more inclusive solution. The research process includes collecting hand gesture datasets using MediaPipe, splitting the dataset into training and testing data with a 75:25 ratio, and training the model using a Learning Rate Scheduler. The model architecture is designed to capture patterns from keypoint data by optimizing the use of dropout layers and the softmax activation function. The evaluation shows that the model achieves an accuracy of 90.22% on the test data, with an average precision of 91%, recall of 89%, and F1-score of 90%. The trial results also demonstrate consistent performance for simple gestures, while accuracy decreases for complex gestures and greater distances. This research provides a significant contribution to enabling voice-free interaction, particularly for deaf individuals, by integrating LSTM technology into interactive kiosk systems.]]>
Copyrights © 2025
