Abstract
We show quantitative evidence that a full relational model of the body performs better at upper body parsing than the standard tree model, despite the need to adopt approximate inference and learning procedures. Our method uses an approximate search for inference, and an approximate structure learning method to learn. We compare our method to state of the art methods on our dataset (which depicts a wide range of poses), on the standard Buffy dataset, and on the reduced PASCAL dataset published recently. Our results suggest that the Buffy dataset over emphasizes poses where the arms hang down, and that leads to generalization problems.
| Original language | English (US) |
|---|---|
| Title of host publication | Computer Vision, ECCV 2010 - 11th European Conference on Computer Vision, Proceedings |
| Publisher | Springer-Verlag |
| Pages | 227-240 |
| Number of pages | 14 |
| Edition | PART 4 |
| ISBN (Print) | 364215560X, 9783642155604 |
| DOIs | |
| State | Published - Jan 1 2010 |
| Event | 11th European Conference on Computer Vision, ECCV 2010 - Heraklion, Crete, Greece Duration: Sep 10 2010 → Sep 11 2010 |
Publication series
| Name | Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) |
|---|---|
| Number | PART 4 |
| Volume | 6314 LNCS |
| ISSN (Print) | 0302-9743 |
| ISSN (Electronic) | 1611-3349 |
Conference
| Conference | 11th European Conference on Computer Vision, ECCV 2010 |
|---|---|
| Country | Greece |
| City | Heraklion, Crete |
| Period | 9/10/10 → 9/11/10 |
Fingerprint
ASJC Scopus subject areas
- Theoretical Computer Science
- Computer Science(all)
Cite this
Improved human parsing with a full relational model. / Tran, Duan; Forsyth, David.
Computer Vision, ECCV 2010 - 11th European Conference on Computer Vision, Proceedings. PART 4. ed. Springer-Verlag, 2010. p. 227-240 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 6314 LNCS, No. PART 4).Research output: Chapter in Book/Report/Conference proceeding › Conference contribution
}
TY - GEN
T1 - Improved human parsing with a full relational model
AU - Tran, Duan
AU - Forsyth, David
PY - 2010/1/1
Y1 - 2010/1/1
N2 - We show quantitative evidence that a full relational model of the body performs better at upper body parsing than the standard tree model, despite the need to adopt approximate inference and learning procedures. Our method uses an approximate search for inference, and an approximate structure learning method to learn. We compare our method to state of the art methods on our dataset (which depicts a wide range of poses), on the standard Buffy dataset, and on the reduced PASCAL dataset published recently. Our results suggest that the Buffy dataset over emphasizes poses where the arms hang down, and that leads to generalization problems.
AB - We show quantitative evidence that a full relational model of the body performs better at upper body parsing than the standard tree model, despite the need to adopt approximate inference and learning procedures. Our method uses an approximate search for inference, and an approximate structure learning method to learn. We compare our method to state of the art methods on our dataset (which depicts a wide range of poses), on the standard Buffy dataset, and on the reduced PASCAL dataset published recently. Our results suggest that the Buffy dataset over emphasizes poses where the arms hang down, and that leads to generalization problems.
UR - http://www.scopus.com/inward/record.url?scp=78149316204&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=78149316204&partnerID=8YFLogxK
U2 - 10.1007/978-3-642-15561-1_17
DO - 10.1007/978-3-642-15561-1_17
M3 - Conference contribution
AN - SCOPUS:78149316204
SN - 364215560X
SN - 9783642155604
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 227
EP - 240
BT - Computer Vision, ECCV 2010 - 11th European Conference on Computer Vision, Proceedings
PB - Springer-Verlag
ER -