Details
| Originalsprache | Englisch |
|---|---|
| Titel des Sammelwerks | Pattern Recognition |
| Untertitel | 39th German Conference, GCPR 2017, Proceedings |
| Herausgeber (Verlag) | Springer Verlag |
| Seiten | 347-360 |
| Seitenumfang | 14 |
| ISBN (Print) | 9783319667089 |
| Publikationsstatus | Veröffentlicht - 15 Aug. 2017 |
| Veranstaltung | 39th German Conference on Pattern Recognition, GCPR 2017 - Basel, Schweiz Dauer: 12 Sept. 2017 → 15 Sept. 2017 |
Publikationsreihe
| Name | Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) |
|---|---|
| Band | 10496 LNCS |
| ISSN (Print) | 0302-9743 |
| ISSN (elektronisch) | 1611-3349 |
Abstract
This paper presents a method to estimate 3D human pose and body shape from monocular videos. While recent approaches infer the 3D pose from silhouettes and landmarks, we exploit properties of optical flow to temporally constrain the reconstructed motion. We estimate human motion by minimizing the difference between computed flow fields and the output of our novel flow renderer. By just using a single semi-automatic initialization step, we are able to reconstruct monocular sequences without joint annotation. Our test scenarios demonstrate that optical flow effectively regularizes the under-constrained problem of human shape and motion estimation from monocular video.
ASJC Scopus Sachgebiete
- Mathematik (insg.)
- Theoretische Informatik
- Informatik (insg.)
- Allgemeine Computerwissenschaft
Zitieren
- Standard
- Harvard
- Apa
- Vancouver
- BibTex
- RIS
Pattern Recognition: 39th German Conference, GCPR 2017, Proceedings. Springer Verlag, 2017. S. 347-360 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Band 10496 LNCS).
Publikation: Beitrag in Buch/Bericht/Sammelwerk/Konferenzband › Aufsatz in Konferenzband › Forschung › Peer-Review
}
TY - GEN
T1 - Optical flow-based 3D human motion estimation from monocular video
AU - Alldieck, Thiemo
AU - Kassubeck, Marc
AU - Wandt, Bastian
AU - Rosenhahn, Bodo
AU - Magnor, Marcus
N1 - Funding information: Acknowledgments. The authors gratefully acknowledge funding by the German Science Foundation from project DFG MA2555/12-1.
PY - 2017/8/15
Y1 - 2017/8/15
N2 - This paper presents a method to estimate 3D human pose and body shape from monocular videos. While recent approaches infer the 3D pose from silhouettes and landmarks, we exploit properties of optical flow to temporally constrain the reconstructed motion. We estimate human motion by minimizing the difference between computed flow fields and the output of our novel flow renderer. By just using a single semi-automatic initialization step, we are able to reconstruct monocular sequences without joint annotation. Our test scenarios demonstrate that optical flow effectively regularizes the under-constrained problem of human shape and motion estimation from monocular video.
AB - This paper presents a method to estimate 3D human pose and body shape from monocular videos. While recent approaches infer the 3D pose from silhouettes and landmarks, we exploit properties of optical flow to temporally constrain the reconstructed motion. We estimate human motion by minimizing the difference between computed flow fields and the output of our novel flow renderer. By just using a single semi-automatic initialization step, we are able to reconstruct monocular sequences without joint annotation. Our test scenarios demonstrate that optical flow effectively regularizes the under-constrained problem of human shape and motion estimation from monocular video.
UR - http://www.scopus.com/inward/record.url?scp=85029573342&partnerID=8YFLogxK
U2 - 10.1007/978-3-319-66709-6_28
DO - 10.1007/978-3-319-66709-6_28
M3 - Conference contribution
AN - SCOPUS:85029573342
SN - 9783319667089
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 347
EP - 360
BT - Pattern Recognition
PB - Springer Verlag
T2 - 39th German Conference on Pattern Recognition, GCPR 2017
Y2 - 12 September 2017 through 15 September 2017
ER -