Evaluation of CNN Models with Transfer Learning for Recognition of Sign Language Alphabets with Complex Background

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

4 Scopus citations

Abstract

Sign language acts as a mediator for deaf and speech-impaired community to visually communicate with one another and engage in their environment. As a state-of-the-art recognition technique in computer vision, deep learning models have demonstrated success for several tasks. In this paper, we present a Convolutional Neural Network (CNN) approach with transfer learning to recognize Arabic and American Sign Languages alphabets with complex background. The descent underlying concept of transfer learning is to adopt a model pretrained on a massive annotated dataset and fine tune the later layers on the target dataset. We applied different techniques to improve the accuracy of the proposed approach such as data augmentation, batch-normalization, and early stopping. The proposed model is evaluated on three datasets and experiments reveal improved results with high recognition rates.

Original languageEnglish
Title of host publication2020 International Conference on Innovation and Intelligence for Informatics, Computing and Technologies, 3ICT 2020
PublisherInstitute of Electrical and Electronics Engineers Inc.
ISBN (Electronic)9781728196732
DOIs
StatePublished - 20 Dec 2020

Publication series

Name2020 International Conference on Innovation and Intelligence for Informatics, Computing and Technologies, 3ICT 2020

Bibliographical note

Publisher Copyright:
© 2020 IEEE.

Keywords

  • Convolutional neural network.
  • Deep learning
  • Gesture recognition
  • Sign language recognition
  • Transfer learning

ASJC Scopus subject areas

  • Artificial Intelligence
  • Computer Science Applications
  • Information Systems

Fingerprint

Dive into the research topics of 'Evaluation of CNN Models with Transfer Learning for Recognition of Sign Language Alphabets with Complex Background'. Together they form a unique fingerprint.

Cite this