EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Comparative Evaluation of LSTM and 3D CNN Models in a Hybrid System for IoT-Enabled Sign-to-Text Translation in Deaf Communities

Samar Mouti, Hani Al Chalabi, Mohammed Abushohada, Samer Rihawi, Sulafa Abdalla

Informatics · 05.02.2026 · 13 (2) · 27

DOI: 10.3390/informatics13020027 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Informatics, online erschienen am 05.02.2026. Verfasst von Samar Mouti und 4 weiteren Autor:innen. Die Arbeit wurde bisher 3-mal zitiert (Crossref) und nennt 28 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Liwa University, United Arab Emirates, through the Faculty Research Incentive Grant, Liwa University.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

This paper presents a hybrid deep learning framework for real-time sign language recognition (SLR) tailored to Internet of Things (IoT)-enabled environments, enhancing accessibility for Deaf communities. The proposed system integrates a Long Short-Term Memory (LSTM) network for static gesture recognition and a 3D Convolutional Neural Network (3D CNN) for dynamic gesture recognition. Implemented on a Raspberry Pi device using MediaPipe for landmark extraction, the system supports low-latency, on-device inference suitable for resource-constrained edge computing. Experimental results demonstrate that the LSTM model achieves its highest stability and performance for static signs at 1000 training epochs, yielding an average F1-score of 0.938 and an accuracy of 86.67%. In contrast, at 2000 epochs, the model exhibits a catastrophic performance collapse (F1-score of 0.088) due to overfitting and weight instability, highlighting the necessity of careful training regulation. Despite this, the overall system achieves consistently high classification performance under controlled conditions. In contrast, the 3D CNN component maintains robust and consistent performance across all evaluated training phases (500–2000 epochs), achieving up to 99.6% accuracy on dynamic signs. When deployed on a Raspberry Pi platform, the system achieves real-time performance with a frame rate of 12–15 FPS and an average inference latency of approximately 65 ms per frame. The hybrid architecture effectively balances recognition accuracy with computational efficiency by routing static gestures to the LSTM and dynamic gestures to the 3D CNN. This work presents a detailed epoch-wise comparative analysis of model stability and computational feasibility, contributing a practical and scalable IoT-enabled solution for inclusive, real-time sign-to-text communication in intelligent environments.

Förderung

  • Liwa University, United Arab Emirates, through the Faculty Research Incentive Grant
  • Liwa University

Zitieren

APA

Mouti, S., Chalabi, H. A., Abushohada, M., Rihawi, S., & Abdalla, S. (2026). Comparative Evaluation of LSTM and 3D CNN Models in a Hybrid System for IoT-Enabled Sign-to-Text Translation in Deaf Communities. Informatics, 13(2), 27. https://doi.org/10.3390/informatics13020027

Vancouver

Mouti S, Chalabi HA, Abushohada M, Rihawi S, Abdalla S. Comparative Evaluation of LSTM and 3D CNN Models in a Hybrid System for IoT-Enabled Sign-to-Text Translation in Deaf Communities. Informatics. 2026;13(2):27. doi:10.3390/informatics13020027

BibTeX
@article{Mouti2026,
  author = {Mouti, Samar and Chalabi, Hani Al and Abushohada, Mohammed and Rihawi, Samer and Abdalla, Sulafa},
  title = {Comparative Evaluation of LSTM and 3D CNN Models in a Hybrid System for IoT-Enabled Sign-to-Text Translation in Deaf Communities},
  journal = {Informatics},
  year = {2026},
  volume = {13},
  number = {2},
  pages = {27},
  doi = {10.3390/informatics13020027}
}
Kostenlos starten →