Evaluating State-of-the-art Object Detector on Challenging Traffic Light Data

Morten Bornø Jensen; Kamal Nasrollahi; Thomas B. Moeslund

doi:10.1109/CVPRW.2017.122

Evaluating State-of-the-art Object Detector on Challenging Traffic Light Data

Morten Bornø Jensen, Kamal Nasrollahi, Thomas B. Moeslund

Research output: Contribution to book/anthology/report/conference proceeding › Article in proceeding › Research › peer-review

66 Citations (Scopus)

229 Downloads (Pure)

Abstract

Traffic light detection (TLD) is a vital part of both intel- ligent vehicles and driving assistance systems (DAS). Gen- eral for most TLDs is that they are evaluated on small and private datasets making it hard to determine the exact per- formance of a given method. In this paper we apply the state-of-the-art, real-time object detection system You Only Look Once, (YOLO) on the public LISA Traffic Light dataset available through the VIVA-challenge, which contain a high number of annotated traffic lights, captured in varying light and weather conditions.
The YOLO object detector achieves an AUC of impres- sively 90.49 % for daysequence1, which is an improvement of 50.32 % compared to the latest ACF entry in the VIVA- challenge. Using the exact same training configuration as the ACF detector, the YOLO detector reaches an AUC of 58.3 %, which is in an increase of 18.13 %.

Original language	English
Title of host publication	2017 IEEE Conference on Computer Vision and Pattern Recognition Workshops : Traffic Surveillance Workshop and Challenge
Publisher	IEEE
Publication date	21 Jul 2017
ISBN (Print)	978-1-5386-0734-3
ISBN (Electronic)	978-1-5386-0733-6
DOIs	https://doi.org/10.1109/CVPRW.2017.122
Publication status	Published - 21 Jul 2017
Event	2017 IEEE Conference on Computer Vision and Pattern Recognition Workshops: Traffic Surveillance Workshop and Challenge - Hawaii Convention Center, Honolulu, United States Duration: 21 Jul 2017 → 26 Jul 2017

Conference

Conference	2017 IEEE Conference on Computer Vision and Pattern Recognition Workshops: Traffic Surveillance Workshop and Challenge
Location	Hawaii Convention Center
Country/Territory	United States
City	Honolulu
Period	21/07/2017 → 26/07/2017

Series	IEEE Conference on Computer Vision and Pattern Recognition Workshops (CVPRW)
ISSN	2160-7516

Access to Document

10.1109/CVPRW.2017.122

Jensen_Evaluating_State-Of-The-Art_Object_CVPR_2017_paperAccepted author manuscript, 0.98 MB

http://openaccess.thecvf.com/content_cvpr_2017_workshops/w9/papers/Jensen_Evaluating_State-Of-The-Art_Object_CVPR_2017_paper.pdf

AUB Link

Search for the material in Aalborg University Library's search engine

Cite this

@inproceedings{d66ada98bc78476aaf487af9f3c2f655,

title = "Evaluating State-of-the-art Object Detector on Challenging Traffic Light Data",

abstract = "Traffic light detection (TLD) is a vital part of both intel- ligent vehicles and driving assistance systems (DAS). Gen- eral for most TLDs is that they are evaluated on small and private datasets making it hard to determine the exact per- formance of a given method. In this paper we apply the state-of-the-art, real-time object detection system You Only Look Once, (YOLO) on the public LISA Traffic Light dataset available through the VIVA-challenge, which contain a high number of annotated traffic lights, captured in varying light and weather conditions.The YOLO object detector achieves an AUC of impres- sively 90.49 % for daysequence1, which is an improvement of 50.32 % compared to the latest ACF entry in the VIVA- challenge. Using the exact same training configuration as the ACF detector, the YOLO detector reaches an AUC of 58.3 %, which is in an increase of 18.13 %.",

author = "Jensen, {Morten Born{\o}} and Kamal Nasrollahi and Moeslund, {Thomas B.}",

year = "2017",

month = jul,

day = "21",

doi = "10.1109/CVPRW.2017.122",

language = "English",

isbn = "978-1-5386-0734-3",

series = "IEEE Conference on Computer Vision and Pattern Recognition Workshops (CVPRW)",

publisher = "IEEE",

booktitle = "2017 IEEE Conference on Computer Vision and Pattern Recognition Workshops",

address = "United States",

note = "2017 IEEE Conference on Computer Vision and Pattern Recognition Workshops: Traffic Surveillance Workshop and Challenge ; Conference date: 21-07-2017 Through 26-07-2017",

}

Jensen, MB, Nasrollahi, K & Moeslund, TB 2017, Evaluating State-of-the-art Object Detector on Challenging Traffic Light Data. in 2017 IEEE Conference on Computer Vision and Pattern Recognition Workshops: Traffic Surveillance Workshop and Challenge. IEEE, IEEE Conference on Computer Vision and Pattern Recognition Workshops (CVPRW), 2017 IEEE Conference on Computer Vision and Pattern Recognition Workshops: Traffic Surveillance Workshop and Challenge, Honolulu, Hawaii, United States, 21/07/2017. https://doi.org/10.1109/CVPRW.2017.122

Evaluating State-of-the-art Object Detector on Challenging Traffic Light Data. / Jensen, Morten Bornø; Nasrollahi, Kamal ; Moeslund, Thomas B.
2017 IEEE Conference on Computer Vision and Pattern Recognition Workshops: Traffic Surveillance Workshop and Challenge. IEEE, 2017. (IEEE Conference on Computer Vision and Pattern Recognition Workshops (CVPRW)).

Research output: Contribution to book/anthology/report/conference proceeding › Article in proceeding › Research › peer-review

TY - GEN

T1 - Evaluating State-of-the-art Object Detector on Challenging Traffic Light Data

AU - Jensen, Morten Bornø

AU - Nasrollahi, Kamal

AU - Moeslund, Thomas B.

PY - 2017/7/21

Y1 - 2017/7/21

N2 - Traffic light detection (TLD) is a vital part of both intel- ligent vehicles and driving assistance systems (DAS). Gen- eral for most TLDs is that they are evaluated on small and private datasets making it hard to determine the exact per- formance of a given method. In this paper we apply the state-of-the-art, real-time object detection system You Only Look Once, (YOLO) on the public LISA Traffic Light dataset available through the VIVA-challenge, which contain a high number of annotated traffic lights, captured in varying light and weather conditions.The YOLO object detector achieves an AUC of impres- sively 90.49 % for daysequence1, which is an improvement of 50.32 % compared to the latest ACF entry in the VIVA- challenge. Using the exact same training configuration as the ACF detector, the YOLO detector reaches an AUC of 58.3 %, which is in an increase of 18.13 %.

AB - Traffic light detection (TLD) is a vital part of both intel- ligent vehicles and driving assistance systems (DAS). Gen- eral for most TLDs is that they are evaluated on small and private datasets making it hard to determine the exact per- formance of a given method. In this paper we apply the state-of-the-art, real-time object detection system You Only Look Once, (YOLO) on the public LISA Traffic Light dataset available through the VIVA-challenge, which contain a high number of annotated traffic lights, captured in varying light and weather conditions.The YOLO object detector achieves an AUC of impres- sively 90.49 % for daysequence1, which is an improvement of 50.32 % compared to the latest ACF entry in the VIVA- challenge. Using the exact same training configuration as the ACF detector, the YOLO detector reaches an AUC of 58.3 %, which is in an increase of 18.13 %.

U2 - 10.1109/CVPRW.2017.122

DO - 10.1109/CVPRW.2017.122

M3 - Article in proceeding

SN - 978-1-5386-0734-3

T3 - IEEE Conference on Computer Vision and Pattern Recognition Workshops (CVPRW)

BT - 2017 IEEE Conference on Computer Vision and Pattern Recognition Workshops

PB - IEEE

T2 - 2017 IEEE Conference on Computer Vision and Pattern Recognition Workshops: Traffic Surveillance Workshop and Challenge

Y2 - 21 July 2017 through 26 July 2017

ER -

Evaluating State-of-the-art Object Detector on Challenging Traffic Light Data

Abstract

Conference

Access to Document

AUB Link

Fingerprint

Cite this