Penerapan Long Short Term Memory (lstm) Untuk Pengenalan Bahasa Isyarat Sederhana Indonesia (bisindo) Secara Realtime (studi Kasus Di Kedai Susu Tuli)

Arrozaqi, David (2025) Penerapan Long Short Term Memory (lstm) Untuk Pengenalan Bahasa Isyarat Sederhana Indonesia (bisindo) Secara Realtime (studi Kasus Di Kedai Susu Tuli). Undergraduate thesis, Politeknik Negeri Jember.

[img] Text (Abstract)
Abstract.pdf - Submitted Version
Available under License Creative Commons Attribution Share Alike.
Download (170kB)
[img] Text (Bab 1 Pendahuluan)
Bab 1 Pendahuluan.pdf - Submitted Version
Available under License Creative Commons Attribution Share Alike.
Download (198kB)
[img] Text (Daftar Pustaka)
Daftar Pustaka.pdf - Submitted Version
Available under License Creative Commons Attribution Share Alike.
Download (162kB)
[img] Text (Laporan Lengkap)
Laporan Lengkap.pdf - Submitted Version
Restricted to Registered users only
Download (3MB) | Request a copy

Abstract

Indonesian Sign Language (BISINDO) is a visual language used by individuals who are deaf and hard of hearing to communicate in daily life. This study aims to develop a web-based application capable of recognizing BISINDO gestures and translating them into speech in real-time. The case study was conducted at Kedai Susu Tuli (K-Suli), where static learning media such as books and posters are less effective for capturing the dynamic movements of sign language. The developed system utilizes MediaPipe Holistic to extract body pose, facial, and hand landmarks from real-time camera input. The extracted data is then processed using a Long Short-Term Memory (LSTM) model with a BiLSTM architecture to classify the gestures. Subsequently, a text-to-speech feature based on gTTS converts the recognized gestures into speech instantly. The dataset used consists of 1,500 videos covering 10 vocabulary words. Testing results show that the application can recognize gestures with an average real-time accuracy of 80% by non-deaf users. Furthermore, User Acceptance Testing (UAT) recorded a user satisfaction rate of 91.4%. Therefore, this system offers an innovative solution to improve accessibility in BISINDO recognition.

Item Type: Thesis (Undergraduate)
Contributors:
ContributionContributorsNIDN/NIDK
Thesis advisorKurniasari, Arvita AgusNIDN0031089301/6163771672230223
Uncontrolled Keywords: : Sign Language, BISINDO, Long Short-Term Memory (LSTM), MediaPipe Holistic, Realtime, Web Application, Text-to-Speech
Subjects: 410 - Rumpun Ilmu Teknik > 450 - Teknik Elektro dan Informatika > 453 - Teknik Telekomunikasi
410 - Rumpun Ilmu Teknik > 450 - Teknik Elektro dan Informatika > 458 - Teknik Informatika
410 - Rumpun Ilmu Teknik > 450 - Teknik Elektro dan Informatika > 461 - Sistem Informasi
410 - Rumpun Ilmu Teknik > 450 - Teknik Elektro dan Informatika > 462 - Teknologi Informasi
410 - Rumpun Ilmu Teknik > 450 - Teknik Elektro dan Informatika > 463 - Teknik Perangkat Lunak
500 - Rumpun Ilmu Bahasa > 510 - Ilmu Sastra (dan Bahasa) Indonesia dan Daerah > 512 - Sastra (dan Bahasa) Indonesia
Divisions: Jurusan Teknologi Informasi > Prodi D4 Teknik Informatika > Tugas Akhir
Depositing User: David Arrozaqi
Date Deposited: 18 Jun 2025 03:03
Last Modified: 18 Jun 2025 03:03
URI: https://sipora.polije.ac.id/id/eprint/41992

Actions (login required)

View Item View Item
Sistem Informasi Polije Repository Asset (SIPORA) is powered by EPrints 3 which is developed by the School of Electronics and Computer Science at the University of Southampton. More information and software credits.