Improving Line Drawing Generation with Instance Segmentation Information 


Vol. 48,  No. 11, pp. 1457-1463, Nov.  2023
10.7840/kics.2023.48.11.1457


PDF
  Abstract

The line drawing generation model aims to generate stylized line drawing images from original photographs using style transfer techniques. Prior work trained semantic and geometric information from photographs to generate line drawing without paired line drawing. It utilized CLIP (contrastive language-image pretraining) for semantic information and employed depth estimation for geometric information. Due to lack of ground-truth depth information, baseline used pseudo-labels instead of ground-truth, showing lower performance than using ground-truth. Therefore, our approach aims to generate high-quality line drawing images by incorporating an object segmentation method that utilizes ground-truth of depth information. By adding object segmentation information, it compensates insufficient ground-truth information for depth estimation and added geometric information such as background and shading. As a result, our approach achieved better performance on the publicly available various image datasets for line drawing.

  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.


  Related Articles
  Cite this article

[IEEE Style]

J. Choi and J. Lee, "Improving Line Drawing Generation with Instance Segmentation Information," The Journal of Korean Institute of Communications and Information Sciences, vol. 48, no. 11, pp. 1457-1463, 2023. DOI: 10.7840/kics.2023.48.11.1457.

[ACM Style]

Jaewoong Choi and Jaekoo Lee. 2023. Improving Line Drawing Generation with Instance Segmentation Information. The Journal of Korean Institute of Communications and Information Sciences, 48, 11, (2023), 1457-1463. DOI: 10.7840/kics.2023.48.11.1457.

[KICS Style]

Jaewoong Choi and Jaekoo Lee, "Improving Line Drawing Generation with Instance Segmentation Information," The Journal of Korean Institute of Communications and Information Sciences, vol. 48, no. 11, pp. 1457-1463, 11. 2023. (https://doi.org/10.7840/kics.2023.48.11.1457)
Vol. 48, No. 11 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