An optimization approach for localization refinement of candidate traffic signs

Zhu, Zhe, Lu, Jiaming, Martin, Ralph Robert and Hu, Shimin 2017. An optimization approach for localization refinement of candidate traffic signs. IEEE Transactions on Intelligent Transportation Systems 10.1109/TITS.2017.2665647

Preview

PDF - Published Version Available under License Creative Commons Attribution. Download (4MB) | Preview

Abstract

We propose a localisation refinement approach for candidate traffic signs. Previous traffic sign localisation approaches which place a bounding rectangle around the sign do not always give a compact bounding box, making the subsequent classification task more difficult. We formulate localisation as a segmentation problem, and incorporate prior knowledge concerning color and shape of traffic signs. To evaluate the effectiveness of our approach, we use it as an intermediate step between a standard traffic sign localizer and a classifier. Our experiments use the well-known GTSDB benchmark as well as our new CTSDB (Chinese Traffic Sign Detection Benchmark). This newly created benchmark is publicly available, and goes beyond previous benchmark datasets: it has over 5,000 highresolution images containing more than 14,000 traffic signs taken in realistic driving conditions. Experimental results show that our localization approach significantly improves bounding boxes when compared to a standard localizer, thereby allowing a standard traffic sign classifier to generate more accurate classification results.

Item Type:	Article
Date Type:	Published Online
Status:	Published
Schools:	Computer Science & Informatics
Subjects:	Q Science > QA Mathematics > QA75 Electronic computers. Computer science
Uncontrolled Keywords:	Shape, Image color analysis, Detectors, Feature extraction, Image segmentation, Benchmark testing, Standards
Publisher:	Institute of Electrical and Electronics Engineers
ISSN:	1524-9050
Funders:	EPSRC
Date of First Compliant Deposit:	14 August 2017
Date of Acceptance:	27 January 2017
Last Modified:	05 May 2023 22:55
URI:	https://orca.cardiff.ac.uk/id/eprint/97836

Citation Data

Cited 20 times in Scopus. View in Scopus. Powered By Scopus® Data

Actions (repository staff only)

Edit Item