Deteksi Kosakata Bahasa Isyarat Indonesia (BISINDO) Menggunakan Deep Learning Berbasis Body Pose

Caesario, Muhammad Abyan Farhan (2023) Deteksi Kosakata Bahasa Isyarat Indonesia (BISINDO) Menggunakan Deep Learning Berbasis Body Pose. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 07211840000027-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 October 2025. Download (11MB) | Request a copy

Abstract

Tunarungu dan tunawicara menggunakan bahasa isyarat untuk dapat berkomunikasi dengan komunitas masyarakat lain. Kemajuan teknologi dapat memudahkan penerjemahan bahasa isyarat dengan tepat dan mudah dimengerti oleh masyarakat pada umumnya. Terdapat dua Bahasa Isyarat di Indonesia yakni SIBI (Sistem Isyarat Bahasa Indonesia) dan BISINDO (Bahasa Isyarat Indonesia) yang keduanya memiliki gerakan yang berbeda. BISINDO sudah lama dikenal dan digunakan oleh masyarakat Indonesia sedangkan SIBI diadopsi dari Bahasa
Isyarat Amerika (ASL). Penelitian ini mengembangkan sistem deteksi kosakata BISINDO menggunakan pendekatan Deep learning dengan Body Pose. Sistem yang dikembangkan yakni mengekstraksi skeleton dari dataset citra yang menggunakan MediaPipe sebagai input dataset dalam model CNN. Hasil penelitian menunjukkan bahwa Model CNN dapat mendeteksi
klasifikasi kosakata BISINDO secara baik dengan tingkat akurasi sebesar 96–99%. Tingkat akurasi Model CNN tersebut lebih baik dibandingkan akurasi Model LSTM yakni 85–97%
=====================================================================================================================================
The deaf and mute use sign language to be able to ommunicate with other communities. Advances in technology can facilitate the translation of sign language accurately and easily understood by the general public. There are two Sign Languages in Indonesia, namely SIBI (Indonesian Signing System) and BISINDO (Indonesian Sign Language), both of which have different movements. BISINDO has long been known and used by Indonesian people while SIBI was adopted from American Sign Language (ASL). This research develops the BISINDO vocabulary detection system using the Deep Learning approach with Body Pose. The developed system is extracting the skeleton from the image dataset using MediaPipe as the dataset input in the CNN model. The results showed that the CNN model could detect BISINDO vocabulary classifications well with an accuracy rate of 96–99%. The level of accuracy of the CNN Model is better than the accuracy of the LSTM Model, which is 85–97%.

Item Type:	Thesis (Other)
Additional Information:	RSKom 004.019 Cae d-1 2023
Uncontrolled Keywords:	BISINDO, Neural Network, Body Pose, Deep Learning, CNN, LSTM
Subjects:	Q Science > Q Science (General) > Q325.5 Machine learning. Support vector machines.
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Computer Engineering > 90243-(S1) Undergraduate Thesis
Depositing User:	Muhammad Abyan Farhan Caesario
Date Deposited:	03 Aug 2023 07:45
Last Modified:	09 Jan 2024 03:12
URI:	http://repository.its.ac.id/id/eprint/101015

Actions (login required)

View Item