No Cover Image

Conference Paper/Proceeding/Abstract 713 views 449 downloads

Detecting humans in RGB-D data with CNNs

Kaiyang Zhou, Adeline Paiement, Majid Mirmehdi

Pages: 306 - 309

Swansea University Author: Adeline Paiement

DOI (Published version): 10.23919/MVA.2017.7986862

Abstract

We address the problem of people detection in RGB-D data where we leverage depth information to develop a region-of-interest (ROI) selection method that provides proposals to two color and depth CNNs. To combine the detections produced by the two CNNs, we propose a novel fusion approach based on the...

Full description

Published: The Fifteenth IAPR International Conference on Machine Vision Applications 2017
Online Access: http://www.mva-org.jp/mva2017/
URI: https://cronfa.swan.ac.uk/Record/cronfa31971
Tags: Add Tag
No Tags, Be the first to tag this record!
first_indexed 2017-02-13T20:56:40Z
last_indexed 2018-02-09T05:19:23Z
id cronfa31971
recordtype SURis
fullrecord <?xml version="1.0"?><rfc1807><datestamp>2017-09-13T15:23:42.2918552</datestamp><bib-version>v2</bib-version><id>31971</id><entry>2017-02-13</entry><title>Detecting humans in RGB-D data with CNNs</title><swanseaauthors><author><sid>f50adf4186d930e3a2a0f9a6d643cf53</sid><firstname>Adeline</firstname><surname>Paiement</surname><name>Adeline Paiement</name><active>true</active><ethesisStudent>false</ethesisStudent></author></swanseaauthors><date>2017-02-13</date><deptcode>FGHSS</deptcode><abstract>We address the problem of people detection in RGB-D data where we leverage depth information to develop a region-of-interest (ROI) selection method that provides proposals to two color and depth CNNs. To combine the detections produced by the two CNNs, we propose a novel fusion approach based on the characteristics of depth images. We also present a new depth-encoding scheme, which not only encodes depth images into three channels but also enhances the information for classification. We conduct experiments on a publicly available RGB-D people dataset and show that our approach outperforms the baseline models that only use RGB data.</abstract><type>Conference Paper/Proceeding/Abstract</type><journal/><paginationStart>306</paginationStart><paginationEnd>309</paginationEnd><publisher>The Fifteenth IAPR International Conference on Machine Vision Applications</publisher><keywords/><publishedDay>20</publishedDay><publishedMonth>7</publishedMonth><publishedYear>2017</publishedYear><publishedDate>2017-07-20</publishedDate><doi>10.23919/MVA.2017.7986862</doi><url>http://www.mva-org.jp/mva2017/</url><notes/><college>COLLEGE NANME</college><department>Humanities and Social Sciences - Faculty</department><CollegeCode>COLLEGE CODE</CollegeCode><DepartmentCode>FGHSS</DepartmentCode><institution>Swansea University</institution><apcterm/><lastEdited>2017-09-13T15:23:42.2918552</lastEdited><Created>2017-02-13T15:08:25.6267803</Created><path><level id="1">Faculty of Science and Engineering</level><level id="2">School of Mathematics and Computer Science - Computer Science</level></path><authors><author><firstname>Kaiyang</firstname><surname>Zhou</surname><order>1</order></author><author><firstname>Adeline</firstname><surname>Paiement</surname><order>2</order></author><author><firstname>Majid</firstname><surname>Mirmehdi</surname><order>3</order></author></authors><documents><document><filename>0031971-02062017183257.pdf</filename><originalFilename>MVA17.pdf</originalFilename><uploaded>2017-06-02T18:32:57.2700000</uploaded><type>Output</type><contentLength>5534229</contentLength><contentType>application/pdf</contentType><version>Version of Record</version><cronfaStatus>true</cronfaStatus><embargoDate>2017-07-20T00:00:00.0000000</embargoDate><copyrightCorrect>true</copyrightCorrect><language>eng</language></document></documents><OutputDurs/></rfc1807>
spelling 2017-09-13T15:23:42.2918552 v2 31971 2017-02-13 Detecting humans in RGB-D data with CNNs f50adf4186d930e3a2a0f9a6d643cf53 Adeline Paiement Adeline Paiement true false 2017-02-13 FGHSS We address the problem of people detection in RGB-D data where we leverage depth information to develop a region-of-interest (ROI) selection method that provides proposals to two color and depth CNNs. To combine the detections produced by the two CNNs, we propose a novel fusion approach based on the characteristics of depth images. We also present a new depth-encoding scheme, which not only encodes depth images into three channels but also enhances the information for classification. We conduct experiments on a publicly available RGB-D people dataset and show that our approach outperforms the baseline models that only use RGB data. Conference Paper/Proceeding/Abstract 306 309 The Fifteenth IAPR International Conference on Machine Vision Applications 20 7 2017 2017-07-20 10.23919/MVA.2017.7986862 http://www.mva-org.jp/mva2017/ COLLEGE NANME Humanities and Social Sciences - Faculty COLLEGE CODE FGHSS Swansea University 2017-09-13T15:23:42.2918552 2017-02-13T15:08:25.6267803 Faculty of Science and Engineering School of Mathematics and Computer Science - Computer Science Kaiyang Zhou 1 Adeline Paiement 2 Majid Mirmehdi 3 0031971-02062017183257.pdf MVA17.pdf 2017-06-02T18:32:57.2700000 Output 5534229 application/pdf Version of Record true 2017-07-20T00:00:00.0000000 true eng
title Detecting humans in RGB-D data with CNNs
spellingShingle Detecting humans in RGB-D data with CNNs
Adeline Paiement
title_short Detecting humans in RGB-D data with CNNs
title_full Detecting humans in RGB-D data with CNNs
title_fullStr Detecting humans in RGB-D data with CNNs
title_full_unstemmed Detecting humans in RGB-D data with CNNs
title_sort Detecting humans in RGB-D data with CNNs
author_id_str_mv f50adf4186d930e3a2a0f9a6d643cf53
author_id_fullname_str_mv f50adf4186d930e3a2a0f9a6d643cf53_***_Adeline Paiement
author Adeline Paiement
author2 Kaiyang Zhou
Adeline Paiement
Majid Mirmehdi
format Conference Paper/Proceeding/Abstract
container_start_page 306
publishDate 2017
institution Swansea University
doi_str_mv 10.23919/MVA.2017.7986862
publisher The Fifteenth IAPR International Conference on Machine Vision Applications
college_str Faculty of Science and Engineering
hierarchytype
hierarchy_top_id facultyofscienceandengineering
hierarchy_top_title Faculty of Science and Engineering
hierarchy_parent_id facultyofscienceandengineering
hierarchy_parent_title Faculty of Science and Engineering
department_str School of Mathematics and Computer Science - Computer Science{{{_:::_}}}Faculty of Science and Engineering{{{_:::_}}}School of Mathematics and Computer Science - Computer Science
url http://www.mva-org.jp/mva2017/
document_store_str 1
active_str 0
description We address the problem of people detection in RGB-D data where we leverage depth information to develop a region-of-interest (ROI) selection method that provides proposals to two color and depth CNNs. To combine the detections produced by the two CNNs, we propose a novel fusion approach based on the characteristics of depth images. We also present a new depth-encoding scheme, which not only encodes depth images into three channels but also enhances the information for classification. We conduct experiments on a publicly available RGB-D people dataset and show that our approach outperforms the baseline models that only use RGB data.
published_date 2017-07-20T03:39:06Z
_version_ 1763751751722205184
score 10.998138

Similar Items