File Download

There are no files associated with this item.

  Links for fulltext
     (May Require Subscription)
Supplementary

Article: A Sampling Approach to Generating Closely Interacting 3D Pose-Pairs from 2D Annotations

TitleA Sampling Approach to Generating Closely Interacting 3D Pose-Pairs from 2D Annotations
Authors
Keywordsdata generation and augmentation
Closely interacting 3D human poses
MCMC sampling
Issue Date2019
Citation
IEEE Transactions on Visualization and Computer Graphics, 2019, v. 25, n. 6, p. 2217-2227 How to Cite?
Abstract© 1995-2012 IEEE. We introduce a data-driven method to generate a large number of plausible, closely interacting 3D human pose-pairs, for a given motion category, e.g., wrestling or salsa dance. With much difficulty in acquiring close interactions using 3D sensors, our approach utilizes abundant existing video data which cover many human activities. Instead of treating the data generation problem as one of reconstruction, either through 3D acquisition or direct 2D-to-3D data lifting from video annotations, we present a solution based on Markov Chain Monte Carlo (MCMC) sampling. Given a motion category and a set of video frames depicting the motion with the 2D pose-pair in each frame annotated, we start the sampling with one or few seed 3D pose-pairs which are manually created based on the target motion category. The initial set is then augmented by MCMC sampling around the seeds, via the Metropolis-Hastings algorithm and guided by a probability density function (PDF) that is defined by two terms to bias the sampling towards 3D pose-pairs that are physically valid and plausible for the motion category. With a focus on efficient sampling over the space of close interactions, rather than pose spaces, we develop a novel representation called interaction coordinates (IC) to encode both poses and their interactions in an integrated manner. Plausibility of a 3D pose-pair is then defined based on the IC and with respect to the annotated 2D pose-pairs from video. We show that our sampling-based approach is able to efficiently synthesize a large volume of plausible, closely interacting 3D pose-pairs which provide a good coverage of the input 2D pose-pairs.
Persistent Identifierhttp://hdl.handle.net/10722/288583
ISSN
2023 Impact Factor: 4.7
2023 SCImago Journal Rankings: 2.056
ISI Accession Number ID

 

DC FieldValueLanguage
dc.contributor.authorYin, Kangxue-
dc.contributor.authorHuang, Hui-
dc.contributor.authorHo, Edmond S.L.-
dc.contributor.authorWang, Hao-
dc.contributor.authorKomura, Taku-
dc.contributor.authorCohen-Or, Daniel-
dc.contributor.authorZhang, Hao-
dc.date.accessioned2020-10-12T08:05:20Z-
dc.date.available2020-10-12T08:05:20Z-
dc.date.issued2019-
dc.identifier.citationIEEE Transactions on Visualization and Computer Graphics, 2019, v. 25, n. 6, p. 2217-2227-
dc.identifier.issn1077-2626-
dc.identifier.urihttp://hdl.handle.net/10722/288583-
dc.description.abstract© 1995-2012 IEEE. We introduce a data-driven method to generate a large number of plausible, closely interacting 3D human pose-pairs, for a given motion category, e.g., wrestling or salsa dance. With much difficulty in acquiring close interactions using 3D sensors, our approach utilizes abundant existing video data which cover many human activities. Instead of treating the data generation problem as one of reconstruction, either through 3D acquisition or direct 2D-to-3D data lifting from video annotations, we present a solution based on Markov Chain Monte Carlo (MCMC) sampling. Given a motion category and a set of video frames depicting the motion with the 2D pose-pair in each frame annotated, we start the sampling with one or few seed 3D pose-pairs which are manually created based on the target motion category. The initial set is then augmented by MCMC sampling around the seeds, via the Metropolis-Hastings algorithm and guided by a probability density function (PDF) that is defined by two terms to bias the sampling towards 3D pose-pairs that are physically valid and plausible for the motion category. With a focus on efficient sampling over the space of close interactions, rather than pose spaces, we develop a novel representation called interaction coordinates (IC) to encode both poses and their interactions in an integrated manner. Plausibility of a 3D pose-pair is then defined based on the IC and with respect to the annotated 2D pose-pairs from video. We show that our sampling-based approach is able to efficiently synthesize a large volume of plausible, closely interacting 3D pose-pairs which provide a good coverage of the input 2D pose-pairs.-
dc.languageeng-
dc.relation.ispartofIEEE Transactions on Visualization and Computer Graphics-
dc.subjectdata generation and augmentation-
dc.subjectClosely interacting 3D human poses-
dc.subjectMCMC sampling-
dc.titleA Sampling Approach to Generating Closely Interacting 3D Pose-Pairs from 2D Annotations-
dc.typeArticle-
dc.description.naturelink_to_subscribed_fulltext-
dc.identifier.doi10.1109/TVCG.2018.2832097-
dc.identifier.pmid29994049-
dc.identifier.scopuseid_2-s2.0-85046363098-
dc.identifier.volume25-
dc.identifier.issue6-
dc.identifier.spage2217-
dc.identifier.epage2227-
dc.identifier.eissn1941-0506-
dc.identifier.isiWOS:000466910200007-
dc.identifier.issnl1077-2626-

Export via OAI-PMH Interface in XML Formats


OR


Export to Other Non-XML Formats