Faster than the fastest : using calibrated cameras to improve the fastest pedestrian detector in the west
Autor(a) principal: | |
---|---|
Data de Publicação: | 2015 |
Tipo de documento: | Trabalho de conclusão de curso |
Idioma: | eng |
Título da fonte: | Repositório Institucional da UFRGS |
Texto Completo: | http://hdl.handle.net/10183/126065 |
Resumo: | In this thesis, we translated a state-of-the-art object detector (the Dóllar method) to the C++ programming language and explored ways to use camera calibration to improve its performance by reducing the amount of calculations necessary and to improve the results by taking away false positives. We developed these techniques in the context of pedestrian detection. On data sets more aligned with video surveillance applications (the camera is high in relation to the ground and far from the area where objects are expected to be), we had great results across the board: the amount of scales in the feature pyramid is reduced by about half, the amount of times the classifier is applied is greatly reduced together with the number of false detections, all while keeping the loss in detection coverage manageable. We also tested our detector in one data set that closely resembles the use of detection in robotics or self-driving systems for automobiles (camera closer to the ground plane and parallel to it). The results suggest the method needs adjustments to be applied to this type of setting. Although there was no loss in detection quality and both the number of scales in the feature pyramid and the number of false positives were reduced, the amount of classifier applications seems excessive. To avoid this problem, we need to adjust the Dense Detection phase of our method (subsection 3.2.2) to account for the fact that images created by these camera settings have a bigger range of possible pedestrian heights and more portions of the image are plausible to provide detections. |
id |
UFRGS-2_432ed0824d7c4e39aa96dd16871ea0df |
---|---|
oai_identifier_str |
oai:www.lume.ufrgs.br:10183/126065 |
network_acronym_str |
UFRGS-2 |
network_name_str |
Repositório Institucional da UFRGS |
repository_id_str |
|
spelling |
Arnoud, Charles Daniel RibeiroJung, Claudio Rosito2015-08-29T02:40:10Z2015http://hdl.handle.net/10183/126065000972318In this thesis, we translated a state-of-the-art object detector (the Dóllar method) to the C++ programming language and explored ways to use camera calibration to improve its performance by reducing the amount of calculations necessary and to improve the results by taking away false positives. We developed these techniques in the context of pedestrian detection. On data sets more aligned with video surveillance applications (the camera is high in relation to the ground and far from the area where objects are expected to be), we had great results across the board: the amount of scales in the feature pyramid is reduced by about half, the amount of times the classifier is applied is greatly reduced together with the number of false detections, all while keeping the loss in detection coverage manageable. We also tested our detector in one data set that closely resembles the use of detection in robotics or self-driving systems for automobiles (camera closer to the ground plane and parallel to it). The results suggest the method needs adjustments to be applied to this type of setting. Although there was no loss in detection quality and both the number of scales in the feature pyramid and the number of false positives were reduced, the amount of classifier applications seems excessive. To avoid this problem, we need to adjust the Dense Detection phase of our method (subsection 3.2.2) to account for the fact that images created by these camera settings have a bigger range of possible pedestrian heights and more portions of the image are plausible to provide detections.application/pdfengComputação gráficaVisualizaçãoComputer visionPedestrian detectionFaster than the fastest : using calibrated cameras to improve the fastest pedestrian detector in the westinfo:eu-repo/semantics/publishedVersioninfo:eu-repo/semantics/bachelorThesisUniversidade Federal do Rio Grande do SulInstituto de InformáticaPorto Alegre, BR-RS2015Ciência da Computação: Ênfase em Ciência da Computação: Bachareladograduaçãoinfo:eu-repo/semantics/openAccessreponame:Repositório Institucional da UFRGSinstname:Universidade Federal do Rio Grande do Sul (UFRGS)instacron:UFRGSORIGINAL000972318.pdf000972318.pdfTexto completo (inglês)application/pdf7113995http://www.lume.ufrgs.br/bitstream/10183/126065/1/000972318.pdfcacb9912517e6b25d35f6a488309fd27MD51TEXT000972318.pdf.txt000972318.pdf.txtExtracted Texttext/plain43081http://www.lume.ufrgs.br/bitstream/10183/126065/2/000972318.pdf.txt8a25c4ef3b3871e7463fbb99759e8d10MD52THUMBNAIL000972318.pdf.jpg000972318.pdf.jpgGenerated Thumbnailimage/jpeg1067http://www.lume.ufrgs.br/bitstream/10183/126065/3/000972318.pdf.jpg7ea144e24ba06ef76268605eb63dd038MD5310183/1260652021-05-07 05:03:37.391632oai:www.lume.ufrgs.br:10183/126065Repositório de PublicaçõesPUBhttps://lume.ufrgs.br/oai/requestopendoar:2021-05-07T08:03:37Repositório Institucional da UFRGS - Universidade Federal do Rio Grande do Sul (UFRGS)false |
dc.title.pt_BR.fl_str_mv |
Faster than the fastest : using calibrated cameras to improve the fastest pedestrian detector in the west |
title |
Faster than the fastest : using calibrated cameras to improve the fastest pedestrian detector in the west |
spellingShingle |
Faster than the fastest : using calibrated cameras to improve the fastest pedestrian detector in the west Arnoud, Charles Daniel Ribeiro Computação gráfica Visualização Computer vision Pedestrian detection |
title_short |
Faster than the fastest : using calibrated cameras to improve the fastest pedestrian detector in the west |
title_full |
Faster than the fastest : using calibrated cameras to improve the fastest pedestrian detector in the west |
title_fullStr |
Faster than the fastest : using calibrated cameras to improve the fastest pedestrian detector in the west |
title_full_unstemmed |
Faster than the fastest : using calibrated cameras to improve the fastest pedestrian detector in the west |
title_sort |
Faster than the fastest : using calibrated cameras to improve the fastest pedestrian detector in the west |
author |
Arnoud, Charles Daniel Ribeiro |
author_facet |
Arnoud, Charles Daniel Ribeiro |
author_role |
author |
dc.contributor.author.fl_str_mv |
Arnoud, Charles Daniel Ribeiro |
dc.contributor.advisor1.fl_str_mv |
Jung, Claudio Rosito |
contributor_str_mv |
Jung, Claudio Rosito |
dc.subject.por.fl_str_mv |
Computação gráfica Visualização |
topic |
Computação gráfica Visualização Computer vision Pedestrian detection |
dc.subject.eng.fl_str_mv |
Computer vision Pedestrian detection |
description |
In this thesis, we translated a state-of-the-art object detector (the Dóllar method) to the C++ programming language and explored ways to use camera calibration to improve its performance by reducing the amount of calculations necessary and to improve the results by taking away false positives. We developed these techniques in the context of pedestrian detection. On data sets more aligned with video surveillance applications (the camera is high in relation to the ground and far from the area where objects are expected to be), we had great results across the board: the amount of scales in the feature pyramid is reduced by about half, the amount of times the classifier is applied is greatly reduced together with the number of false detections, all while keeping the loss in detection coverage manageable. We also tested our detector in one data set that closely resembles the use of detection in robotics or self-driving systems for automobiles (camera closer to the ground plane and parallel to it). The results suggest the method needs adjustments to be applied to this type of setting. Although there was no loss in detection quality and both the number of scales in the feature pyramid and the number of false positives were reduced, the amount of classifier applications seems excessive. To avoid this problem, we need to adjust the Dense Detection phase of our method (subsection 3.2.2) to account for the fact that images created by these camera settings have a bigger range of possible pedestrian heights and more portions of the image are plausible to provide detections. |
publishDate |
2015 |
dc.date.accessioned.fl_str_mv |
2015-08-29T02:40:10Z |
dc.date.issued.fl_str_mv |
2015 |
dc.type.status.fl_str_mv |
info:eu-repo/semantics/publishedVersion |
dc.type.driver.fl_str_mv |
info:eu-repo/semantics/bachelorThesis |
format |
bachelorThesis |
status_str |
publishedVersion |
dc.identifier.uri.fl_str_mv |
http://hdl.handle.net/10183/126065 |
dc.identifier.nrb.pt_BR.fl_str_mv |
000972318 |
url |
http://hdl.handle.net/10183/126065 |
identifier_str_mv |
000972318 |
dc.language.iso.fl_str_mv |
eng |
language |
eng |
dc.rights.driver.fl_str_mv |
info:eu-repo/semantics/openAccess |
eu_rights_str_mv |
openAccess |
dc.format.none.fl_str_mv |
application/pdf |
dc.source.none.fl_str_mv |
reponame:Repositório Institucional da UFRGS instname:Universidade Federal do Rio Grande do Sul (UFRGS) instacron:UFRGS |
instname_str |
Universidade Federal do Rio Grande do Sul (UFRGS) |
instacron_str |
UFRGS |
institution |
UFRGS |
reponame_str |
Repositório Institucional da UFRGS |
collection |
Repositório Institucional da UFRGS |
bitstream.url.fl_str_mv |
http://www.lume.ufrgs.br/bitstream/10183/126065/1/000972318.pdf http://www.lume.ufrgs.br/bitstream/10183/126065/2/000972318.pdf.txt http://www.lume.ufrgs.br/bitstream/10183/126065/3/000972318.pdf.jpg |
bitstream.checksum.fl_str_mv |
cacb9912517e6b25d35f6a488309fd27 8a25c4ef3b3871e7463fbb99759e8d10 7ea144e24ba06ef76268605eb63dd038 |
bitstream.checksumAlgorithm.fl_str_mv |
MD5 MD5 MD5 |
repository.name.fl_str_mv |
Repositório Institucional da UFRGS - Universidade Federal do Rio Grande do Sul (UFRGS) |
repository.mail.fl_str_mv |
|
_version_ |
1801224488878080000 |