A Deep-Learning-Based Method for Recognizing Existence of Power-Lines in Infrared Images 


Vol. 45,  No. 1, pp. 159-162, Jan.  2020
10.7840/kics.2020.45.1.159


PDF Full-Text
  Abstract

In this paper, we propose a deep-learning-based method for recognizing the existence of power lines in infrared images. Deep learning has the advantage of learning feature vectors from a large number of data, resulting in higher performance than conventional methods using hand-crafted feature vectors in various domains such as image recognition and object detection. Taking this advantage, we propose a method based on deep learning, which identifies the presence of power lines in infrared images. In order to find the most appropriate architecture, we compare five architectures based on VGGNet and ResNet. As a result, the proposed method achieves an accuracy of 98.65%, which is better than the state-of-the-art DCT-based method.

  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]

J. Kim, S. Shin, C. Jung, C. Kim, "A Deep-Learning-Based Method for Recognizing Existence of Power-Lines in Infrared Images," The Journal of Korean Institute of Communications and Information Sciences, vol. 45, no. 1, pp. 159-162, 2020. DOI: 10.7840/kics.2020.45.1.159.

[ACM Style]

Jonghee Kim, Seungkwon Shin, Chanho Jung, and Changick Kim. 2020. A Deep-Learning-Based Method for Recognizing Existence of Power-Lines in Infrared Images. The Journal of Korean Institute of Communications and Information Sciences, 45, 1, (2020), 159-162. DOI: 10.7840/kics.2020.45.1.159.

[KICS Style]

Jonghee Kim, Seungkwon Shin, Chanho Jung, Changick Kim, "A Deep-Learning-Based Method for Recognizing Existence of Power-Lines in Infrared Images," The Journal of Korean Institute of Communications and Information Sciences, vol. 45, no. 1, pp. 159-162, 1. 2020. (https://doi.org/10.7840/kics.2020.45.1.159)

Search
(IN TITLE, AUTHOR, ABSTRACT,KEYWORDS)

Advanced Search ▼

POPULAR KEYWORDS
(TOP 10 KEYWORDS)

Recent Publications
(LAST 3 YEARS)






pISSN : 1226 - 4717
eISSN : 2287 - 3880