PointMotionNet: Point-Wise Motion Learning for Large-Scale LiDAR Point Clouds Sequences

Simple item page
dc.contributor.authorWang, Junen
dc.contributor.authorLi, Xiaolongen
dc.contributor.authorSullivan, Alanen
dc.contributor.authorAbbott, A. Lynnen
dc.contributor.authorChen, Sihengen
dc.date.accessioned2023-02-28T15:40:55Zen
dc.date.available2023-02-28T15:40:55Zen
dc.date.issued2022-06en
dc.date.updated2023-02-25T19:32:36Zen
dc.description.abstractWe propose a point-based spatiotemporal pyramid architecture, called PointMotionNet, to learn motion information from a sequence of large-scale 3D LiDAR point clouds. A core component of PointMotionNet is a novel technique for point-based spatiotemporal convolution, which finds the point correspondences across time by leveraging a time-invariant spatial neighboring space and extracts spatiotemporal features. To validate PointMotionNet, we consider two motion-related tasks: point-based motion prediction and multisweep semantic segmentation. For each task, we design an end-to-end system where PointMotionNet is the core module that learns motion information. We conduct extensive experiments and show that i) for point-based motion prediction, PointMotionNet achieves less than 0.5m mean squared error on Argoverse dataset, which is a significant improvement over existing methods; and ii) for multisweep semantic segmentation, PointMotionNet with a pretrained segmentation backbone outperforms previous SOTA by over 3.3 % mIoU on SemanticKITTI dataset with 25 classes including 6 moving objects.en
dc.description.versionAccepted versionen
dc.format.extentPages 4418-4427en
dc.format.mimetypeapplication/pdfen
dc.identifier.doihttps://doi.org/10.1109/CVPRW56347.2022.00488en
dc.identifier.eissn2160-7516en
dc.identifier.isbn9781665487399en
dc.identifier.issn2160-7508en
dc.identifier.orcidAbbott, Amos [0000-0003-3850-6771]en
dc.identifier.urihttp://hdl.handle.net/10919/114009en
dc.identifier.volume2022-Juneen
dc.language.isoenen
dc.publisherIEEEen
dc.rightsIn Copyrighten
dc.rights.urihttp://rightsstatements.org/vocab/InC/1.0/en
dc.titlePointMotionNet: Point-Wise Motion Learning for Large-Scale LiDAR Point Clouds Sequencesen
dc.title.serialIEEE Computer Society Conference on Computer Vision and Pattern Recognition Workshopsen
dc.typeConference proceedingen
dc.type.dcmitypeTexten
dc.type.otherConference Proceedingen
pubs.finish-date2022-06-20en
pubs.organisational-group/Virginia Techen
pubs.organisational-group/Virginia Tech/Engineeringen
pubs.organisational-group/Virginia Tech/Engineering/Electrical and Computer Engineeringen
pubs.organisational-group/Virginia Tech/All T&R Facultyen
pubs.organisational-group/Virginia Tech/Engineering/COE T&R Facultyen
pubs.start-date2022-06-19en

Files

Original bundle
Now showing 1 - 1 of 1
Thumbnail Image
Name:
Wang_PointMotionNet_Point-Wise_Motion_Learning_for_Large-Scale_LiDAR_Point_Clouds_Sequences_CVPRW_2022_paper.pdf
Size:
9.3 MB
Format:
Adobe Portable Document Format
Description:
Accepted version
Download

Collections

All Faculty Deposits
Scholarly Works, Electrical and Computer Engineering