The Development of a Sign Language Detection System Using TensorFlow and CVzone

Dinda Putri Ramadani (Universitas Dinamika Bangsa)
Prayitno (Department of Informatics, Dinamika Bangsa University, Jambi, Indonesia)
Praditya Oktanza Djaduk Wibisono (Department of Informatics, Dinamika Bangsa University, Jambi, Indonesia)
Mochammad Rizky Abadi Sutoyo (Study Program of Electrical Engineering, Faculty of Science and Technology, Universitas Jambi, Jambi, Indonesia)

Media Journal of General Computer Science (MJGCS)
Vol. 3 No. 2 (2026): MJGCS

Sign language is a communication system used to help deaf people interact with each other. In Indonesia, there are two main sign languages, namely Indonesian Sign Language (BISINDO) and Indonesian Sign Language System (SIBI), each of which has differences in the use of hands. This research develops a SIBI-based sign language recognition system using TensorFlow and CVzone. This system uses computer vision technology and machine learning to recognize SIBI alphabet hand gestures. The model is developed using a dataset captured by a camera and trained with TensorFlow to enable real-time gesture recognition, while CVzone facilitates effective gesture tracking. Test results show that the average accuracy in recognizing the SIBI alphabet reaches more than 90%, which reflects the system's good performance in recognizing the alphabet. It is expected that the accuracy of detecting hand gestures in sign language will be improved to support communication for the deaf community in Indonesia.