Rethinking Real-Time Lane Detection Technology for Autonomous Driving 


Vol. 48,  No. 5, pp. 589-599, May  2023
10.7840/kics.2023.48.5.589


PDF
  Abstract

In autonomous driving, the lane detection system for road driving is one of the essential functions not only for fully autonomous driving but also for LKS (Lane Keeping System) in ADAS (Advanced Driving Assistance System). However, such a lane detection system must operate accurately and in real time under various conditions on the road (optical reflection, weather, night time, occlusion). In this paper, a deep learning-based lane detection system that can operate in real time even at night in the Korean road environment is considered so that strategies for performance optimization are presented. To this end, in addition to the TuSimple data set, which is the foreign terrain daytime training dataset, a Korean terrain daytime and nighttime dataset was newly built and data augmentation techniques were added for training. Moreover the accuracy and real-time operation speed were measured on a single NVIDIA 2080ti GPU. As a result of this additional dataset construction and data augmentation additional training, accuracy improved from 76.9% to 92.4% accuracy at night time in Korea based on the Ultra Fast Lane Detection model, and from 84.6% accuracy to 96.7% accuracy based on the CLRNet model could make an improvement.

  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]

DaewonKwak, JisangYoo, MinjunSon, MinsuPark, DonggeonChoi, SungjinLee, "Rethinking Real-Time Lane Detection Technology for Autonomous Driving," The Journal of Korean Institute of Communications and Information Sciences, vol. 48, no. 5, pp. 589-599, 2023. DOI: 10.7840/kics.2023.48.5.589.

[ACM Style]

DaewonKwak, JisangYoo, MinjunSon, MinsuPark, DonggeonChoi, and SungjinLee. 2023. Rethinking Real-Time Lane Detection Technology for Autonomous Driving. The Journal of Korean Institute of Communications and Information Sciences, 48, 5, (2023), 589-599. DOI: 10.7840/kics.2023.48.5.589.

[KICS Style]

DaewonKwak, JisangYoo, MinjunSon, MinsuPark, DonggeonChoi, SungjinLee, "Rethinking Real-Time Lane Detection Technology for Autonomous Driving," The Journal of Korean Institute of Communications and Information Sciences, vol. 48, no. 5, pp. 589-599, 5. 2023. (https://doi.org/10.7840/kics.2023.48.5.589)
Vol. 48, No. 5 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