<div class="csl-bib-body">
<div class="csl-entry">Windbacher, F., Hödlmoser, M., & Gelautz, M. (2023). Single-Stage 3D Pose Estimation of Vulnerable Road Users Using Pseudo-Labels. In R. Gade, M. Felsberg, & J.-K. Kämäräinen (Eds.), <i>Image Analysis. 22nd Scandinavian Conference, SCIA 2023, Sirkka, Finland, April 18–21, 2023, Proceedings, Part II</i> (pp. 401–417). Springer. https://doi.org/10.1007/978-3-031-31438-4_27</div>
</div>
-
dc.identifier.uri
http://hdl.handle.net/20.500.12708/188289
-
dc.description.abstract
Human pose estimation of vulnerable road users is an important perception task for autonomous vehicles which can be exploited for intention prediction in order to guide the vehicle’s actions. Single-stage human pose estimation approaches with their potential in terms of simplicity and efficiency have shown only mediocre results in 2D, and have hardly been investigated in 3D in the autonomous driving domain so far. We tackle this challenge with the 2D single-stage human pose estimator KAPAO. We find that KAPAO achieves state-of-the-art performance in our evaluation on domain-specific 2D benchmark datasets, which motivates its extension for application in 3D. To overcome a lack of ground truth vulnerable road user data for 3D pose estimation, we first extend the Waymo Open Dataset with additional 3D pseudo-labels. We create more than one million 3D poses, that we estimate using the dataset’s exhaustive person bounding boxes and associated LiDAR point clouds. Evaluating their quality, we report a mean per joint position error of less than 10 cm. Having access to large-scale domain-specific 3D pose data, we propose a 3D variant of KAPAO that additionally predicts the depths of joints. We evaluate it on our extended Waymo Open Dataset and compare its performance to that of a LiDAR uplifting baseline. The proposed approach is low-latency and produces plausible poses but struggles to estimate absolute depth precisely, particularly at large distances. We alleviate that limitation by implementing a conditional LiDAR-based depth correction.
en
dc.description.sponsorship
FFG - Österr. Forschungsförderungs- gesellschaft mbH
-
dc.language.iso
en
-
dc.subject
computer vision
en
dc.subject
machine learning
en
dc.subject
human pose estimation
en
dc.subject
autonomous driving
en
dc.subject
RGB images
en
dc.subject
LiDAR
en
dc.subject
evaluation
en
dc.title
Single-Stage 3D Pose Estimation of Vulnerable Road Users Using Pseudo-Labels
en
dc.type
Inproceedings
en
dc.type
Konferenzbeitrag
de
dc.contributor.affiliation
emotion3D GmbH, Austria
-
dc.contributor.affiliation
emotion3D GmbH, Austria
-
dc.contributor.editoraffiliation
Aalborg University, Denmark
-
dc.contributor.editoraffiliation
Linköping University, Sweden
-
dc.contributor.editoraffiliation
Tampere University, Finland
-
dc.relation.isbn
978-3-031-31437-7
-
dc.relation.doi
10.1007/978-3-031-31438-4
-
dc.relation.issn
0302-9743
-
dc.description.startpage
401
-
dc.description.endpage
417
-
dc.relation.grantno
879642
-
dc.rights.holder
The Authors, under exclusive license to Springer Nature Switzerland AG
-
dc.type.category
Full-Paper Contribution
-
dc.relation.eissn
1611-3349
-
tuw.booktitle
Image Analysis. 22nd Scandinavian Conference, SCIA 2023, Sirkka, Finland, April 18–21, 2023, Proceedings, Part II
-
tuw.container.volume
13886
-
tuw.peerreviewed
true
-
tuw.book.ispartofseries
Lecture Notes in Computer Science
-
tuw.relation.publisher
Springer
-
tuw.relation.publisherplace
Cham
-
tuw.project.title
Multimodales Sensor-Lichtsystem zum Schutz von verletzlichen Verkehrsteilnehmern
-
tuw.researchTopic.id
I5
-
tuw.researchTopic.name
Visual Computing and Human-Centered Technology
-
tuw.researchTopic.value
100
-
tuw.publication.orgunit
E193-01 - Forschungsbereich Computer Vision
-
tuw.publication.orgunit
E193 - Institut für Visual Computing and Human-Centered Technology
-
tuw.publisher.doi
10.1007/978-3-031-31438-4_27
-
dc.description.numberOfPages
17
-
tuw.author.orcid
0000-0002-9476-0865
-
tuw.editor.orcid
0000-0002-8016-2426
-
tuw.editor.orcid
0000-0002-6096-3648
-
tuw.editor.orcid
0000-0002-5801-4371
-
tuw.event.name
SCIA 2023. Scandinavian Conference on Image Analysis
en
tuw.event.startdate
18-04-2023
-
tuw.event.enddate
21-04-2023
-
tuw.event.online
On Site
-
tuw.event.type
Event for scientific audience
-
tuw.event.place
Sirkka
-
tuw.event.country
FI
-
tuw.event.presenter
Windbacher, Fabian
-
wb.sciencebranch
Informatik
-
wb.sciencebranch.oefos
1020
-
wb.sciencebranch.value
100
-
item.openairecristype
http://purl.org/coar/resource_type/c_5794
-
item.openairetype
conference paper
-
item.cerifentitytype
Publications
-
item.fulltext
no Fulltext
-
item.languageiso639-1
en
-
item.grantfulltext
restricted
-
crisitem.project.funder
FFG - Österr. Forschungsförderungs- gesellschaft mbH
-
crisitem.project.grantno
879642
-
crisitem.author.dept
E649-03 - Fachbereich .digital office
-
crisitem.author.dept
E183 - Institut für Rechnergestützte Automation
-
crisitem.author.dept
E193-01 - Forschungsbereich Computer Vision
-
crisitem.author.orcid
0000-0002-9476-0865
-
crisitem.author.parentorg
E649 - Services Vizerektorat Digitalisierung und Infrastruktur
-
crisitem.author.parentorg
E180 - Fakultät für Informatik
-
crisitem.author.parentorg
E193 - Institut für Visual Computing and Human-Centered Technology