Exploiting temporal and spatial constraints in traffic sign detection from a moving vehicle

Sinisa Segvic, Karla Brkic, Zoran Kalafatic, Axel Pinz

Research output: Contribution to journalArticleResearchpeer-review

Abstract

This paper addresses detection, tracking and recognition of traffic signs in video. Previous research has shown that very good detection recalls can be ob-
tained by state-of-the-art detection algorithms. Unfortunately, satisfactory precision and localization accuracy are more difficultly achieved. We follow the intuitive notion that it should be easier to accurately detect an object from an image sequence than from a single image. We propose a novel two-stage technique which achieves improved detection results by applying temporal and spatial constraints to the occurrences of traffic signs in video. The first stage produces well-aligned temporally consistent detection tracks, by managing
many competing track hypotheses at once. The second stage improves the precision by filtering the detection tracks by a learned discriminative model. The two stages have been evaluated in extensive experiments performed on videos acquired from a moving vehicle. The obtained experimental results clearly confirm the advantages of the proposed technique.
Original languageEnglish
Pages (from-to)649–665
JournalMachine vision and applications
Volume25
Issue number3
DOIs
Publication statusPublished - 2014

Fingerprint

Traffic signs
Experiments

Keywords

  • Video analysis
  • Object detection
  • object tracking
  • discriminative models
  • supervised learning

Fields of Expertise

  • Information, Communication & Computing

Treatment code (Nähere Zuordnung)

  • Basic - Fundamental (Grundlagenforschung)
  • Application

Cite this

Exploiting temporal and spatial constraints in traffic sign detection from a moving vehicle. / Segvic, Sinisa; Brkic, Karla; Kalafatic, Zoran; Pinz, Axel.

In: Machine vision and applications, Vol. 25, No. 3, 2014, p. 649–665.

Research output: Contribution to journalArticleResearchpeer-review

Segvic, Sinisa ; Brkic, Karla ; Kalafatic, Zoran ; Pinz, Axel. / Exploiting temporal and spatial constraints in traffic sign detection from a moving vehicle. In: Machine vision and applications. 2014 ; Vol. 25, No. 3. pp. 649–665.
@article{b9c39a5fe4c641378c50c128af735f43,
title = "Exploiting temporal and spatial constraints in traffic sign detection from a moving vehicle",
abstract = "This paper addresses detection, tracking and recognition of traffic signs in video. Previous research has shown that very good detection recalls can be ob-tained by state-of-the-art detection algorithms. Unfortunately, satisfactory precision and localization accuracy are more difficultly achieved. We follow the intuitive notion that it should be easier to accurately detect an object from an image sequence than from a single image. We propose a novel two-stage technique which achieves improved detection results by applying temporal and spatial constraints to the occurrences of traffic signs in video. The first stage produces well-aligned temporally consistent detection tracks, by managingmany competing track hypotheses at once. The second stage improves the precision by filtering the detection tracks by a learned discriminative model. The two stages have been evaluated in extensive experiments performed on videos acquired from a moving vehicle. The obtained experimental results clearly confirm the advantages of the proposed technique.",
keywords = "Video analysis, Object detection, object tracking, discriminative models, supervised learning",
author = "Sinisa Segvic and Karla Brkic and Zoran Kalafatic and Axel Pinz",
year = "2014",
doi = "10.1007/s00138-011-0396-y",
language = "English",
volume = "25",
pages = "649–665",
journal = "Machine vision and applications",
issn = "0932-8092",
publisher = "Springer Verlag",
number = "3",

}

TY - JOUR

T1 - Exploiting temporal and spatial constraints in traffic sign detection from a moving vehicle

AU - Segvic, Sinisa

AU - Brkic, Karla

AU - Kalafatic, Zoran

AU - Pinz, Axel

PY - 2014

Y1 - 2014

N2 - This paper addresses detection, tracking and recognition of traffic signs in video. Previous research has shown that very good detection recalls can be ob-tained by state-of-the-art detection algorithms. Unfortunately, satisfactory precision and localization accuracy are more difficultly achieved. We follow the intuitive notion that it should be easier to accurately detect an object from an image sequence than from a single image. We propose a novel two-stage technique which achieves improved detection results by applying temporal and spatial constraints to the occurrences of traffic signs in video. The first stage produces well-aligned temporally consistent detection tracks, by managingmany competing track hypotheses at once. The second stage improves the precision by filtering the detection tracks by a learned discriminative model. The two stages have been evaluated in extensive experiments performed on videos acquired from a moving vehicle. The obtained experimental results clearly confirm the advantages of the proposed technique.

AB - This paper addresses detection, tracking and recognition of traffic signs in video. Previous research has shown that very good detection recalls can be ob-tained by state-of-the-art detection algorithms. Unfortunately, satisfactory precision and localization accuracy are more difficultly achieved. We follow the intuitive notion that it should be easier to accurately detect an object from an image sequence than from a single image. We propose a novel two-stage technique which achieves improved detection results by applying temporal and spatial constraints to the occurrences of traffic signs in video. The first stage produces well-aligned temporally consistent detection tracks, by managingmany competing track hypotheses at once. The second stage improves the precision by filtering the detection tracks by a learned discriminative model. The two stages have been evaluated in extensive experiments performed on videos acquired from a moving vehicle. The obtained experimental results clearly confirm the advantages of the proposed technique.

KW - Video analysis

KW - Object detection

KW - object tracking

KW - discriminative models

KW - supervised learning

U2 - 10.1007/s00138-011-0396-y

DO - 10.1007/s00138-011-0396-y

M3 - Article

VL - 25

SP - 649

EP - 665

JO - Machine vision and applications

JF - Machine vision and applications

SN - 0932-8092

IS - 3

ER -