Neural Object Learning for 6D Pose Estimation Using a Few Cluttered Images

Park, K; Patten, T; Vincze, M

Neural Object Learning for 6D Pose Estimation Using a Few Cluttered Images

Park, K Patten, T

Vincze, M

Permalink

Publisher:: Springer
Publication Type:: Conference Proceeding
Citation:: Computer Vision – ECCV 2020, 2020, 12349 LNCS, pp. 656-673
Issue Date:: 2020-01-01

Closed Access

	Filename	Description	Size
	Park2020_Chapter_NeuralObjectLearningFor6DPoseE.pdf	Published version	2.78 MB	Adobe PDF	View/Open

Copyright Clearance Process

Recently Added
In Progress
Closed Access

This item is closed access and not available.

Full metadata record

Field	Value	Language
dc.contributor.author	Park, K
dc.contributor.author	Patten, T https://orcid.org/0000-0003-1139-9451
dc.contributor.author	Vincze, M
dc.date	2020-08-23
dc.date.accessioned	2021-05-16T21:38:03Z
dc.date.available	2021-05-16T21:38:03Z
dc.date.issued	2020-01-01
dc.identifier.citation	Computer Vision – ECCV 2020, 2020, 12349 LNCS, pp. 656-673
dc.identifier.isbn	9783030585471
dc.identifier.issn	0302-9743
dc.identifier.issn	1611-3349
dc.identifier.uri	http://hdl.handle.net/10453/148915
dc.description.abstract	© 2020, Springer Nature Switzerland AG. Recent methods for 6D pose estimation of objects assume either textured 3D models or real images that cover the entire range of target poses. However, it is difficult to obtain textured 3D models and annotate the poses of objects in real scenarios. This paper proposes a method, Neural Object Learning (NOL), that creates synthetic images of objects in arbitrary poses by combining only a few observations from cluttered images. A novel refinement step is proposed to align inaccurate poses of objects in source images, which results in better quality images. Evaluations performed on two public datasets show that the rendered images created by NOL lead to state-of-the-art performance in comparison to methods that use 13 times the number of real images. Evaluations on our new dataset show multiple objects can be trained and recognized simultaneously using a sequence of a fixed scene.
dc.language	en
dc.publisher	Springer
dc.relation.ispartof	Computer Vision – ECCV 2020
dc.relation.ispartof	European Conference on Computer Vision
dc.relation.ispartofseries	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
dc.relation.isbasedon	10.1007/978-3-030-58548-8_38
dc.rights	info:eu-repo/semantics/closedAccess
dc.subject.classification	Artificial Intelligence & Image Processing
dc.title	Neural Object Learning for 6D Pose Estimation Using a Few Cluttered Images
dc.type	Conference Proceeding
utslib.citation.volume	12349 LNCS
utslib.location.activity	Glasgow, UK
pubs.organisational-group	/University of Technology Sydney
pubs.organisational-group	/University of Technology Sydney/Faculty of Engineering and Information Technology
pubs.organisational-group	/University of Technology Sydney/Faculty of Engineering and Information Technology/School of Mechanical and Mechatronic Engineering
utslib.copyright.status	closed_access	*
pubs.consider-herdc	false
dc.date.updated	2021-05-16T21:38:00Z
pubs.finish-date	2020-08-28
pubs.place-of-publication	Switzerland
pubs.publication-status	Published
pubs.start-date	2020-08-23
pubs.volume	12349 LNCS
dc.location	Switzerland

Abstract:

© 2020, Springer Nature Switzerland AG. Recent methods for 6D pose estimation of objects assume either textured 3D models or real images that cover the entire range of target poses. However, it is difficult to obtain textured 3D models and annotate the poses of objects in real scenarios. This paper proposes a method, Neural Object Learning (NOL), that creates synthetic images of objects in arbitrary poses by combining only a few observations from cluttered images. A novel refinement step is proposed to align inaccurate poses of objects in source images, which results in better quality images. Evaluations performed on two public datasets show that the rendered images created by NOL lead to state-of-the-art performance in comparison to methods that use 13 times the number of real images. Evaluations on our new dataset show multiple objects can be trained and recognized simultaneously using a sequence of a fixed scene.

Please use this identifier to cite or link to this item:

http://hdl.handle.net/10453/148915