Sematic Image Segmentation Using Coarse Label Map 


Vol. 44,  No. 9, pp. 1690-1693, Sep.  2019
10.7840/kics.2019.44.9.1690


PDF
  Abstract

Deep learning based semantic segmentation methods require fine label maps, which costs expensive to generate, for training. To solve this problem, we propose a semantic segmentation method using coarse label maps easily obtained. The proposed method defines heat maps of labels from the coarse lable map utilized as unary potential in CRF for semantic segmentation. We used the Cityscape dataset for performance evaluation. Compared to a coarse label map, the proposed method achieved the improved performance. Also our method reaches 66% of the quality of a deep learning based method. We believe that this study guides for improving semantic segmentation and generating fine label maps.

  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. Eun, C. Jung, C. Kim, "Sematic Image Segmentation Using Coarse Label Map," The Journal of Korean Institute of Communications and Information Sciences, vol. 44, no. 9, pp. 1690-1693, 2019. DOI: 10.7840/kics.2019.44.9.1690.

[ACM Style]

Hyunjun Eun, Chanho Jung, and Changick Kim. 2019. Sematic Image Segmentation Using Coarse Label Map. The Journal of Korean Institute of Communications and Information Sciences, 44, 9, (2019), 1690-1693. DOI: 10.7840/kics.2019.44.9.1690.

[KICS Style]

Hyunjun Eun, Chanho Jung, Changick Kim, "Sematic Image Segmentation Using Coarse Label Map," The Journal of Korean Institute of Communications and Information Sciences, vol. 44, no. 9, pp. 1690-1693, 9. 2019. (https://doi.org/10.7840/kics.2019.44.9.1690)

Search
(IN TITLE, AUTHOR, ABSTRACT,KEYWORDS)

Advanced Search ▼

POPULAR KEYWORDS
(TOP 10 KEYWORDS)

Recent Publications
(LAST 3 YEARS)






pISSN : 1226 - 4717
eISSN : 2287 - 3880