Sign Language Recognition with Multi-Stream Neural Network Using Joint Point Image Patches 


Vol. 48,  No. 6, pp. 669-676, Jun.  2023
10.7840/kics.2023.48.6.669


PDF
  Abstract

Recently, many studies on sign language recognition using deep learning and machine learning algorithms have been conducted. Due to the nature of sign language requiring various types of information, various methods have been proposed for word-level sign language recognition. Sign language recognition techniques are divided into image-based methods and pose-based methods. In the beginning, methods using image-based convolutional neural networks came out and showed satisfactory performance. Image-based methods mainly focused on the overall information of sign language. After that, a graph convolutional neural network using pose-based joint information, which was widely used in the field of action recognition, was applied. At this time, most of the information obtained was focused only on relationships in sign language. However, the information obtained with a single-stream neural network was insufficient to capture sign language features that required various types of information, and many papers have been published on sign language recognition through multi-stream neural networks. In this paper, we propose a multi-stream network using joint point image patches, which are data types to obtain face and hand regional information, and a transformer network.

  Statistics
Cumulative Counts from November, 2022
Multiple requests among the same browser session are counted as one view. If you mouse over a chart, the values of data points will be shown.


  Cite this article

[IEEE Style]

H. S. Kang and K. Park, "Sign Language Recognition with Multi-Stream Neural Network Using Joint Point Image Patches," The Journal of Korean Institute of Communications and Information Sciences, vol. 48, no. 6, pp. 669-676, 2023. DOI: 10.7840/kics.2023.48.6.669.

[ACM Style]

Hyeon Seok Kang and Kwang-Hyun Park. 2023. Sign Language Recognition with Multi-Stream Neural Network Using Joint Point Image Patches. The Journal of Korean Institute of Communications and Information Sciences, 48, 6, (2023), 669-676. DOI: 10.7840/kics.2023.48.6.669.

[KICS Style]

Hyeon Seok Kang and Kwang-Hyun Park, "Sign Language Recognition with Multi-Stream Neural Network Using Joint Point Image Patches," The Journal of Korean Institute of Communications and Information Sciences, vol. 48, no. 6, pp. 669-676, 6. 2023. (https://doi.org/10.7840/kics.2023.48.6.669)
Vol. 48, No. 6 Index

Search
(IN TITLE, AUTHOR, ABSTRACT,KEYWORDS)

Advanced Search ▼

POPULAR KEYWORDS
(TOP 10 KEYWORDS)

Recent Publications
(LAST 3 YEARS)






pISSN : 1226 - 4717
eISSN : 2287 - 3880