Implementation of Character Recognizer for Heterogeneous Character Set 


Vol. 46,  No. 8, pp. 1284-1293, Aug.  2021
10.7840/kics.2021.46.8.1284


PDF
  Abstract

In this paper, we present the implementation process of the character recognition engine that inputs hand-written images where four types of characters, Hangul, English, numbers, and special symbols, were mixed. Big data sets for training deep neural network models were prepared through data augmentation techniques based on font-based character images. The CNN-based deep neural network model, the core part of the character recognition engine, was trained on a desktop with a high-performance GPU. The recognition accuracy of the trained neural network model showed a performance of Top-1 accuracy of 0.98. The model trained on the desktop was ported to a lightweight neural network model that includes only inference capabilities, so that it could be executed on an embedded system of relatively limited computing resources. In the embedded system, the trained model was utilized as a character recognition engine in application programs for individual character recognition and word recognition in a simple form of table.

  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. Kim and J. Kim, "Implementation of Character Recognizer for Heterogeneous Character Set," The Journal of Korean Institute of Communications and Information Sciences, vol. 46, no. 8, pp. 1284-1293, 2021. DOI: 10.7840/kics.2021.46.8.1284.

[ACM Style]

Hongsoog Kim and Jeong-Si Kim. 2021. Implementation of Character Recognizer for Heterogeneous Character Set. The Journal of Korean Institute of Communications and Information Sciences, 46, 8, (2021), 1284-1293. DOI: 10.7840/kics.2021.46.8.1284.

[KICS Style]

Hongsoog Kim and Jeong-Si Kim, "Implementation of Character Recognizer for Heterogeneous Character Set," The Journal of Korean Institute of Communications and Information Sciences, vol. 46, no. 8, pp. 1284-1293, 8. 2021. (https://doi.org/10.7840/kics.2021.46.8.1284)

Search
(IN TITLE, AUTHOR, ABSTRACT,KEYWORDS)

Advanced Search ▼

POPULAR KEYWORDS
(TOP 10 KEYWORDS)

Recent Publications
(LAST 3 YEARS)






pISSN : 1226 - 4717
eISSN : 2287 - 3880