Faiz Firstian Nugroho, . (2026) APLIKASI PENDETEKSI BAHASA ISYARAT INDONESIA PADA GESTURE TANGAN MENGGUNAKAN CNN DAN LSTM BERBASIS MOBILE. Skripsi thesis, Universitas Pembangunan Nasional Veteran Jakarta.
![[img]](http://repository.upnvj.ac.id/style/images/fileicons/text.png) |
Text
ABSTRAK.pdf Download (250kB) |
|
Text
AWAL.pdf Download (911kB) |
|
Text
BAB 1.pdf Restricted to Repository UPNVJ Only Download (318kB) |
|
Text
BAB 2.pdf Restricted to Repository UPNVJ Only Download (686kB) |
|
Text
BAB 3.pdf Restricted to Repository UPNVJ Only Download (937kB) |
|
Text
BAB 4.pdf Restricted to Repository UPNVJ Only Download (1MB) |
|
Text
BAB 5.pdf Download (271kB) |
|
Text
DAFTAR PUSTAKA.pdf Download (286kB) |
|
Text
RIWAYAT HIDUP.pdf Restricted to Repository staff only Download (257kB) |
|
Text
LAMPIRAN.pdf Restricted to Repository UPNVJ Only Download (1MB) |
|
Text
HASIL PLAGIARISME.pdf Restricted to Repository staff only Download (20MB) |
|
Text
ARTIKEL KI.pdf Restricted to Repository staff only Download (824kB) |
Abstract
Indonesian Sign Language (BISINDO) serves as the primary means of communication for individuals who are deaf or hard of hearing. However, it is not yet widely understood by the general public, which often leads to barriers in everyday communication.. Previous studies have applied CNN and LSTM in sign language recognition, such as Altiarika and Sari (2023) using CNN and LSTM and Utomo et al. (2024) using MediaPipe and LSTM, but the integration of MobileNetV2 visual features, hand landmarks, and LSTM-based temporal modeling for BISINDO word recognition in a mobile application is still rarely developed. This study aims to develop a native Android mobile application for detecting BISINDO hand gestures as an initial communication aid between the general public and deaf individuals. The method combines a Convolutional Neural Network (CNN) with the MobileNetV2 architecture as a visual feature extractor and Long Short-Term Memory (LSTM) to learn temporal movement patterns. The dataset was collected independently as 255 hand-gesture videos from 15 BISINDO word classes. The development stages include data preprocessing, hand detection using MediaPipe, Landmark Mask and Region of Interest (ROI) formation, visual and landmark feature extraction, sequence formation, training, and model evaluation. The model was then converted to TensorFlow Lite and integrated into an Android application named Sindoraku. Evaluation on the test data achieved 96.01% accuracy, with a macro average precision, recall, and F1-Score of 96% each. The application was evaluated through Black Box Testing and the result all features worked as designed, performance testing and the result real-time detection with an average latency of 249 ms per frame or 3.1 fps and 258 MB memory usage, and User Acceptance Testing with 15 respondents, which obtained an acceptance index of 90.97% with Very Feasible category. Therefore, the application is feasible as a learning aid for basic BISINDO gesture recognition.
| Item Type: | Thesis (Skripsi) |
|---|---|
| Additional Information: | [No.Panggil: 2210511045] [Pembimbing 1: Muhammad Adrezo] [Pembimbing 2: Kharisma Wiati Gusti] [Penguji 1: Neny Rosmawarni] [Penguji 2: Nurul Afifah Arifuddin] |
| Uncontrolled Keywords: | CNN, Indonesian Sign language, LSTM, MobileNetV2, Mobile. |
| Subjects: | Q Science > QA Mathematics > QA75 Electronic computers. Computer science Q Science > QA Mathematics > QA76 Computer software T Technology > T Technology (General) |
| Divisions: | Fakultas Ilmu Komputer > Program Studi Informatika (S1) |
| Depositing User: | FAIZ FIRSTIAN NUGROHO |
| Date Deposited: | 28 Jul 2026 16:00 |
| Last Modified: | 28 Aug 2026 08:01 |
| URI: | http://repository.upnvj.ac.id/id/eprint/51988 |
Actions (login required)
 |
View Item |