Visual people counting using gender features and LRU updating scheme

Chen Chiung Hsieh, Mansour Karkoub, Wei Ru Lai, Po Hong Lin

Research output: Contribution to journalArticle

Abstract

The general public spends a significant amount of time in front of digital signage seeking information from many venues such as exhibition halls and shopping centers. This is why advertisement purchasers believe that the number of passing viewers provides crucial information for their marketing strategies. In this paper, a real-time person counting/memorizing system is designed capable of distinguishing the gender of potential customers. An adaptive boosting (Adaboost) machine learning algorithm is used to detect human faces and utilize specific filtering criteria to eliminate useless data. For each detected person, face and torso information are recorded in a database for identification. The least recently used identification record will be deleted if the database is full. Gender classification is performed by support vector machine using hair ratios extracted from gender characterizing regions. Based on a variety of experiments, the accuracy of the proposed algorithm is higher than 90 % for person-counting and higher than 94 % for gender classification. Moreover, the execution speed on personal computers may reach 15–20 fps.

Original languageEnglish
Pages (from-to)1741-1759
Number of pages19
JournalMultimedia Tools and Applications
Volume74
Issue number6
DOIs
Publication statusPublished - 2015

Fingerprint

Shopping centers
Adaptive boosting
Personal computers
Learning algorithms
Support vector machines
Learning systems
Marketing
Experiments

Keywords

  • Computer vision
  • Digital signage
  • Gender classification
  • People counting

ASJC Scopus subject areas

  • Software
  • Media Technology
  • Hardware and Architecture
  • Computer Networks and Communications

Cite this

Visual people counting using gender features and LRU updating scheme. / Hsieh, Chen Chiung; Karkoub, Mansour; Lai, Wei Ru; Lin, Po Hong.

In: Multimedia Tools and Applications, Vol. 74, No. 6, 2015, p. 1741-1759.

Research output: Contribution to journalArticle

Hsieh, Chen Chiung ; Karkoub, Mansour ; Lai, Wei Ru ; Lin, Po Hong. / Visual people counting using gender features and LRU updating scheme. In: Multimedia Tools and Applications. 2015 ; Vol. 74, No. 6. pp. 1741-1759.
@article{5f9e922b21dd4901a17cbdddfc69423b,
title = "Visual people counting using gender features and LRU updating scheme",
abstract = "The general public spends a significant amount of time in front of digital signage seeking information from many venues such as exhibition halls and shopping centers. This is why advertisement purchasers believe that the number of passing viewers provides crucial information for their marketing strategies. In this paper, a real-time person counting/memorizing system is designed capable of distinguishing the gender of potential customers. An adaptive boosting (Adaboost) machine learning algorithm is used to detect human faces and utilize specific filtering criteria to eliminate useless data. For each detected person, face and torso information are recorded in a database for identification. The least recently used identification record will be deleted if the database is full. Gender classification is performed by support vector machine using hair ratios extracted from gender characterizing regions. Based on a variety of experiments, the accuracy of the proposed algorithm is higher than 90 {\%} for person-counting and higher than 94 {\%} for gender classification. Moreover, the execution speed on personal computers may reach 15–20 fps.",
keywords = "Computer vision, Digital signage, Gender classification, People counting",
author = "Hsieh, {Chen Chiung} and Mansour Karkoub and Lai, {Wei Ru} and Lin, {Po Hong}",
year = "2015",
doi = "10.1007/s11042-013-1715-2",
language = "English",
volume = "74",
pages = "1741--1759",
journal = "Multimedia Tools and Applications",
issn = "1380-7501",
publisher = "Springer Netherlands",
number = "6",

}

TY - JOUR

T1 - Visual people counting using gender features and LRU updating scheme

AU - Hsieh, Chen Chiung

AU - Karkoub, Mansour

AU - Lai, Wei Ru

AU - Lin, Po Hong

PY - 2015

Y1 - 2015

N2 - The general public spends a significant amount of time in front of digital signage seeking information from many venues such as exhibition halls and shopping centers. This is why advertisement purchasers believe that the number of passing viewers provides crucial information for their marketing strategies. In this paper, a real-time person counting/memorizing system is designed capable of distinguishing the gender of potential customers. An adaptive boosting (Adaboost) machine learning algorithm is used to detect human faces and utilize specific filtering criteria to eliminate useless data. For each detected person, face and torso information are recorded in a database for identification. The least recently used identification record will be deleted if the database is full. Gender classification is performed by support vector machine using hair ratios extracted from gender characterizing regions. Based on a variety of experiments, the accuracy of the proposed algorithm is higher than 90 % for person-counting and higher than 94 % for gender classification. Moreover, the execution speed on personal computers may reach 15–20 fps.

AB - The general public spends a significant amount of time in front of digital signage seeking information from many venues such as exhibition halls and shopping centers. This is why advertisement purchasers believe that the number of passing viewers provides crucial information for their marketing strategies. In this paper, a real-time person counting/memorizing system is designed capable of distinguishing the gender of potential customers. An adaptive boosting (Adaboost) machine learning algorithm is used to detect human faces and utilize specific filtering criteria to eliminate useless data. For each detected person, face and torso information are recorded in a database for identification. The least recently used identification record will be deleted if the database is full. Gender classification is performed by support vector machine using hair ratios extracted from gender characterizing regions. Based on a variety of experiments, the accuracy of the proposed algorithm is higher than 90 % for person-counting and higher than 94 % for gender classification. Moreover, the execution speed on personal computers may reach 15–20 fps.

KW - Computer vision

KW - Digital signage

KW - Gender classification

KW - People counting

UR - http://www.scopus.com/inward/record.url?scp=84925292036&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84925292036&partnerID=8YFLogxK

U2 - 10.1007/s11042-013-1715-2

DO - 10.1007/s11042-013-1715-2

M3 - Article

VL - 74

SP - 1741

EP - 1759

JO - Multimedia Tools and Applications

JF - Multimedia Tools and Applications

SN - 1380-7501

IS - 6

ER -