Refining the Performance of Neural Networks with Simple Architectures for Indonesian Sign Language System (SIBI) Letter Recognition Using Keypoint Detection

The diversity of non-verbal communication styles among persons with disabilities in Indonesia highlights the urgent need for technological solutions that support accessibility in both workplace settings and social contexts. This study proposes a novel approach to improving neural network performance...

Full description

Saved in:
Bibliographic Details
Main Authors: Nur Hikma Amir, Chandra Kusuma Dewa, Ahmad Luthfi
Format: Article
Language:English
Published: Fakultas Ilmu Komputer UMI 2025-04-01
Series:Ilkom Jurnal Ilmiah
Subjects:
Online Access:https://jurnal.fikom.umi.ac.id/index.php/ILKOM/article/view/2522
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849414814611275776
author Nur Hikma Amir
Chandra Kusuma Dewa
Ahmad Luthfi
author_facet Nur Hikma Amir
Chandra Kusuma Dewa
Ahmad Luthfi
author_sort Nur Hikma Amir
collection DOAJ
description The diversity of non-verbal communication styles among persons with disabilities in Indonesia highlights the urgent need for technological solutions that support accessibility in both workplace settings and social contexts. This study proposes a novel approach to improving neural network performance through the use of simple architectures for recognizing Indonesian Sign Language (SIBI) letters M and N, by applying keypoint detection while accounting for hand size variations (17–22 cm). Four models were evaluated: YOLOv5 based on image detection, as well as VGG-16, Attention, and Multi-Layer Perceptron (MLP) developed using keypoint detection. The evaluation was conducted in real-time, taking into account accessories such as rings, watches, and gloves, as well as varying lighting intensities to simulate real-world user environments. The novelty lies in the integration of keypoint detection into lightweight architectures, which significantly improves accuracy and resilience against visual disturbances (noise). The MLP model achieved the best performance, with an accuracy of 94% for M and 93% for N, outperforming more complex approaches such as YOLOv5, which showed a significant drop in accuracy under disturbed conditions. The integration of VGG-16 with Attention resulted in underfitting, emphasizing that complexity does not always correlate with effectiveness. These findings underscore the potential of lightweight models to enhance technological accessibility for the disabled community across various social and professional domains.
format Article
id doaj-art-faccef4adecb43b491925c6c150e89c2
institution Kabale University
issn 2087-1716
2548-7779
language English
publishDate 2025-04-01
publisher Fakultas Ilmu Komputer UMI
record_format Article
series Ilkom Jurnal Ilmiah
spelling doaj-art-faccef4adecb43b491925c6c150e89c22025-08-20T03:33:42ZengFakultas Ilmu Komputer UMIIlkom Jurnal Ilmiah2087-17162548-77792025-04-01171647310.33096/ilkom.v17i1.2522.64-73760Refining the Performance of Neural Networks with Simple Architectures for Indonesian Sign Language System (SIBI) Letter Recognition Using Keypoint DetectionNur Hikma Amir0Chandra Kusuma Dewa1Ahmad Luthfi2Universitas Islam IndonesiaUniversitas Islam IndonesiaUniversitas Islam IndonesiaThe diversity of non-verbal communication styles among persons with disabilities in Indonesia highlights the urgent need for technological solutions that support accessibility in both workplace settings and social contexts. This study proposes a novel approach to improving neural network performance through the use of simple architectures for recognizing Indonesian Sign Language (SIBI) letters M and N, by applying keypoint detection while accounting for hand size variations (17–22 cm). Four models were evaluated: YOLOv5 based on image detection, as well as VGG-16, Attention, and Multi-Layer Perceptron (MLP) developed using keypoint detection. The evaluation was conducted in real-time, taking into account accessories such as rings, watches, and gloves, as well as varying lighting intensities to simulate real-world user environments. The novelty lies in the integration of keypoint detection into lightweight architectures, which significantly improves accuracy and resilience against visual disturbances (noise). The MLP model achieved the best performance, with an accuracy of 94% for M and 93% for N, outperforming more complex approaches such as YOLOv5, which showed a significant drop in accuracy under disturbed conditions. The integration of VGG-16 with Attention resulted in underfitting, emphasizing that complexity does not always correlate with effectiveness. These findings underscore the potential of lightweight models to enhance technological accessibility for the disabled community across various social and professional domains.https://jurnal.fikom.umi.ac.id/index.php/ILKOM/article/view/2522disabilitykeypoint detectionletter m and nsibisign language.
spellingShingle Nur Hikma Amir
Chandra Kusuma Dewa
Ahmad Luthfi
Refining the Performance of Neural Networks with Simple Architectures for Indonesian Sign Language System (SIBI) Letter Recognition Using Keypoint Detection
Ilkom Jurnal Ilmiah
disability
keypoint detection
letter m and n
sibi
sign language.
title Refining the Performance of Neural Networks with Simple Architectures for Indonesian Sign Language System (SIBI) Letter Recognition Using Keypoint Detection
title_full Refining the Performance of Neural Networks with Simple Architectures for Indonesian Sign Language System (SIBI) Letter Recognition Using Keypoint Detection
title_fullStr Refining the Performance of Neural Networks with Simple Architectures for Indonesian Sign Language System (SIBI) Letter Recognition Using Keypoint Detection
title_full_unstemmed Refining the Performance of Neural Networks with Simple Architectures for Indonesian Sign Language System (SIBI) Letter Recognition Using Keypoint Detection
title_short Refining the Performance of Neural Networks with Simple Architectures for Indonesian Sign Language System (SIBI) Letter Recognition Using Keypoint Detection
title_sort refining the performance of neural networks with simple architectures for indonesian sign language system sibi letter recognition using keypoint detection
topic disability
keypoint detection
letter m and n
sibi
sign language.
url https://jurnal.fikom.umi.ac.id/index.php/ILKOM/article/view/2522
work_keys_str_mv AT nurhikmaamir refiningtheperformanceofneuralnetworkswithsimplearchitecturesforindonesiansignlanguagesystemsibiletterrecognitionusingkeypointdetection
AT chandrakusumadewa refiningtheperformanceofneuralnetworkswithsimplearchitecturesforindonesiansignlanguagesystemsibiletterrecognitionusingkeypointdetection
AT ahmadluthfi refiningtheperformanceofneuralnetworkswithsimplearchitecturesforindonesiansignlanguagesystemsibiletterrecognitionusingkeypointdetection