Two-Stage Part-Based Pedestrian Detection

Andreas Møgelmose; Antonio Prioletti; Mohan M. Trivedi; Alberto Broggi; Thomas B. Moeslund

doi:10.1109/ITSC.2012.6338898

Two-Stage Part-Based Pedestrian Detection

Andreas Møgelmose, Antonio Prioletti, Mohan M. Trivedi, Alberto Broggi, Thomas B. Moeslund

Publikation: Bidrag til bog/antologi/rapport/konference proceeding › Konferenceartikel i proceeding › Forskning › peer review

19 Citationer (Scopus)

854 Downloads (Pure)

Abstract

This paper introduces a part-based two-stage pedestrian detector. The system finds pedestrian candidates with an AdaBoost cascade on Haar-like features. It then verifies each candidate using a part-based HOG-SVM doing first a regression and then a classification based on the estimated function output from the regression. It uses the Histogram of Oriented Gradients (HOG) computed on both the full, upper and lower body of the candidates, and uses these in the final verification. The system has been trained and tested on the INRIA dataset and performs better than similar previous work, which uses full-body verification.

Originalsprog	Engelsk
Titel	15th International Conference on Intelligent Transportation Systems
Forlag	IEEE
Publikationsdato	16 sep. 2012
Sider	73 - 77
ISBN (Trykt)	978-1-4673-3064-0
DOI	https://doi.org/10.1109/ITSC.2012.6338898
Status	Udgivet - 16 sep. 2012
Begivenhed	Intelligent Transportation Systems Conference - Anchorage, USA Varighed: 16 sep. 2012 → 19 sep. 2012 Konferencens nummer: 15

Konference

Konference	Intelligent Transportation Systems Conference
Nummer	15
Land/Område	USA
By	Anchorage
Periode	16/09/2012 → 19/09/2012

Adgang til dokumentet

10.1109/ITSC.2012.6338898

Two-Stage Part-Based Pedestrian DetectionForlagets udgivne version, 1,68 MB

AUB Link

Søg efter materialet i Aalborg Universitetsbiblioteks søgemaskine

Andre filer og links

Link to publication in Scopus

Citationsformater

@inproceedings{3520eb7c17524caa8357b4bbb11afee6,

title = "Two-Stage Part-Based Pedestrian Detection",

abstract = "Detecting pedestrians is still a challenging task for automotive vision systemdue the extreme variability of targets, lighting conditions, occlusions, andhigh speed vehicle motion. A lot of research has been focused on this problemin the last 10 years and detectors based on classifiers has gained a specialplace among the different approaches presented. This work presents astate-of-the-art pedestrian detection system based on a two stages classifier.Candidates are extracted with a Haar cascade classifier trained with theDaimlerDB dataset and then validated through part-based HOG classifier with the aim of lowering the number of false positives. The surviving candidatesare then filtered with a feature-based tracking to enhance the recognition robustness and improve the results' stability. The system has been implemented on a prototype vehicle and offers high performance in terms ofseveral metrics, such as detection rate, false positives per hour, and framerate. The novelty of this system rely in the combination of HOG part-based approach, tracking based on specific optimized feature and porting on a real prototype.",

author = "Andreas M{\o}gelmose and Antonio Prioletti and Trivedi, {Mohan M.} and Alberto Broggi and Moeslund, {Thomas B.}",

year = "2012",

month = sep,

day = "16",

doi = "10.1109/ITSC.2012.6338898",

language = "English",

isbn = "978-1-4673-3064-0",

pages = "73 -- 77",

booktitle = "15th International Conference on Intelligent Transportation Systems",

publisher = "IEEE",

address = "United States",

note = "Intelligent Transportation Systems Conference ; Conference date: 16-09-2012 Through 19-09-2012",

}

TY - GEN

T1 - Two-Stage Part-Based Pedestrian Detection

AU - Møgelmose, Andreas

AU - Prioletti, Antonio

AU - Trivedi, Mohan M.

AU - Broggi, Alberto

AU - Moeslund, Thomas B.

N1 - Conference code: 15

PY - 2012/9/16

Y1 - 2012/9/16

N2 - Detecting pedestrians is still a challenging task for automotive vision systemdue the extreme variability of targets, lighting conditions, occlusions, andhigh speed vehicle motion. A lot of research has been focused on this problemin the last 10 years and detectors based on classifiers has gained a specialplace among the different approaches presented. This work presents astate-of-the-art pedestrian detection system based on a two stages classifier.Candidates are extracted with a Haar cascade classifier trained with theDaimlerDB dataset and then validated through part-based HOG classifier with the aim of lowering the number of false positives. The surviving candidatesare then filtered with a feature-based tracking to enhance the recognition robustness and improve the results' stability. The system has been implemented on a prototype vehicle and offers high performance in terms ofseveral metrics, such as detection rate, false positives per hour, and framerate. The novelty of this system rely in the combination of HOG part-based approach, tracking based on specific optimized feature and porting on a real prototype.

AB - Detecting pedestrians is still a challenging task for automotive vision systemdue the extreme variability of targets, lighting conditions, occlusions, andhigh speed vehicle motion. A lot of research has been focused on this problemin the last 10 years and detectors based on classifiers has gained a specialplace among the different approaches presented. This work presents astate-of-the-art pedestrian detection system based on a two stages classifier.Candidates are extracted with a Haar cascade classifier trained with theDaimlerDB dataset and then validated through part-based HOG classifier with the aim of lowering the number of false positives. The surviving candidatesare then filtered with a feature-based tracking to enhance the recognition robustness and improve the results' stability. The system has been implemented on a prototype vehicle and offers high performance in terms ofseveral metrics, such as detection rate, false positives per hour, and framerate. The novelty of this system rely in the combination of HOG part-based approach, tracking based on specific optimized feature and porting on a real prototype.

UR - http://www.scopus.com/inward/record.url?scp=84871227060&partnerID=8YFLogxK

U2 - 10.1109/ITSC.2012.6338898

DO - 10.1109/ITSC.2012.6338898

M3 - Article in proceeding

SN - 978-1-4673-3064-0

SP - 73

EP - 77

BT - 15th International Conference on Intelligent Transportation Systems

PB - IEEE

T2 - Intelligent Transportation Systems Conference

Y2 - 16 September 2012 through 19 September 2012

ER -

Two-Stage Part-Based Pedestrian Detection

Abstract

Konference

Adgang til dokumentet

AUB Link

Andre filer og links

Fingeraftryk

Citationsformater