<![CDATA[People with hearing and speech impairments rely on sign language for daily communication. However, public understanding of the Indonesian Sign Language System (SIBI), particularly its alphabet hand gestures, remains limited, creating communication barriers between deaf individuals and the wider community. This study aims to develop an Android-based application for recognizing SIBI alphabet hand gestures using the Transfer Learning method implemented through Google Teachable Machine. A dataset consisting of 200 images for each SIBI alphabet gesture was collected and used to train the classification model. Several training scenarios were evaluated by varying the Epoch, Batch Size, and Learning Rate parameters to obtain the optimal model. The trained model was then converted into TensorFlow Lite and integrated into an Android application for real-time hand gesture recognition. Experimental results show that the best performance was achieved using Epoch 50, Batch Size 16, and Learning Rate 0.001, producing training and validation accuracy of 1.0 with an error rate close to 0.0. The developed application successfully recognized all tested static SIBI alphabet gestures, demonstrating reliable performance in practical implementation. This study contributes by providing a lightweight Android-based SIBI hand gesture recognition system and experimentally evaluating the influence of training hyperparameters on recognition performance. The proposed approach offers an efficient solution for supporting SIBI learning and improving communication accessibility for deaf or speech-impaired individuals.]]>
