Implementasi Bahasa Isyarat Indonesia (Bisindo) Menggunakan Algoritma Convolution Neural Network Dan Pendekatan Transfer Learning Berbasis Android

Deaf and speech-impaired individuals rely on sign language as their primary means of communication. Unfortunately, the general public's limited understanding of sign language remains a barrier to daily interactions. To address this issue, this study developed an Indonesian Sign Language (BISINDO) translator system on Android using the Convolutional Neural Network (CNN) method with a Transfer Learning approach. The system is designed to recognize and translate hand gestures into text in real time. The model was trained using an annotated video dataset and tested on 660 test samples, achieving an accuracy of 80%. This result indicates that the Transfer Learning approach is quite effective in classifying hand gestures. However, challenges remain, particularly with gestures that are visually similar and with limited data coverage. To improve system performance, enrichment of the dataset in both quantity and variation is necessary. The system currently works one-way, translating from sign language to text, and does not yet support reverse translation. Nevertheless, the testing results show that this system can be an effective solution to help the general public interact more easily and efficiently with deaf and speech-impaired individuals.