Easy minimax estimation with random forests for human pose estimation

P. Daphne Tsatsoulis, David Forsyth

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

We describe a method for human parsing that is straightforward and competes with state-of-the-art performance on standard datasets. Unlike the state-of-the-art, our method does not search for individual body parts or poselets. Instead, a regression forest is used to predict a body configuration in body-space. The output of this regression forest is then combined in a novel way. Instead of averaging the output of each tree in the forest we use minimax to calculate optimal weights for the trees. This optimal weighting improves performance on rare poses and improves the generalization of our method to different datasets. Our paper demonstrates the unique advantage of random forest representations: minimax estimation is straightforward with no significant retraining burden.

Original languageEnglish (US)
Title of host publicationComputer Vision - ECCV 2014 Workshops, Proceedings
EditorsMichael M. Bronstein, Carsten Rother, Lourdes Agapito
PublisherSpringer-Verlag
Pages669-684
Number of pages16
ISBN (Electronic)9783319161778
DOIs
StatePublished - Jan 1 2015
Event13th European Conference on Computer Vision, ECCV 2014 - Zurich, Switzerland
Duration: Sep 6 2014Sep 12 2014

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume8925
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Other

Other13th European Conference on Computer Vision, ECCV 2014
CountrySwitzerland
CityZurich
Period9/6/149/12/14

Fingerprint

Minimax Estimation
Pose Estimation
Random Forest
Regression
Output
Parsing
Minimax
Weighting
Averaging
Calculate
Predict
Configuration
Demonstrate
Human

Keywords

  • Human pose estimation
  • Minimax
  • Regression
  • Regression forests

ASJC Scopus subject areas

  • Theoretical Computer Science
  • Computer Science(all)

Cite this

Daphne Tsatsoulis, P., & Forsyth, D. (2015). Easy minimax estimation with random forests for human pose estimation. In M. M. Bronstein, C. Rother, & L. Agapito (Eds.), Computer Vision - ECCV 2014 Workshops, Proceedings (pp. 669-684). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 8925). Springer-Verlag. https://doi.org/10.1007/978-3-319-16178-5_47

Easy minimax estimation with random forests for human pose estimation. / Daphne Tsatsoulis, P.; Forsyth, David.

Computer Vision - ECCV 2014 Workshops, Proceedings. ed. / Michael M. Bronstein; Carsten Rother; Lourdes Agapito. Springer-Verlag, 2015. p. 669-684 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 8925).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Daphne Tsatsoulis, P & Forsyth, D 2015, Easy minimax estimation with random forests for human pose estimation. in MM Bronstein, C Rother & L Agapito (eds), Computer Vision - ECCV 2014 Workshops, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 8925, Springer-Verlag, pp. 669-684, 13th European Conference on Computer Vision, ECCV 2014, Zurich, Switzerland, 9/6/14. https://doi.org/10.1007/978-3-319-16178-5_47
Daphne Tsatsoulis P, Forsyth D. Easy minimax estimation with random forests for human pose estimation. In Bronstein MM, Rother C, Agapito L, editors, Computer Vision - ECCV 2014 Workshops, Proceedings. Springer-Verlag. 2015. p. 669-684. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). https://doi.org/10.1007/978-3-319-16178-5_47
Daphne Tsatsoulis, P. ; Forsyth, David. / Easy minimax estimation with random forests for human pose estimation. Computer Vision - ECCV 2014 Workshops, Proceedings. editor / Michael M. Bronstein ; Carsten Rother ; Lourdes Agapito. Springer-Verlag, 2015. pp. 669-684 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
@inproceedings{2790ab840d4647bbb5882ce5fdcbef72,
title = "Easy minimax estimation with random forests for human pose estimation",
abstract = "We describe a method for human parsing that is straightforward and competes with state-of-the-art performance on standard datasets. Unlike the state-of-the-art, our method does not search for individual body parts or poselets. Instead, a regression forest is used to predict a body configuration in body-space. The output of this regression forest is then combined in a novel way. Instead of averaging the output of each tree in the forest we use minimax to calculate optimal weights for the trees. This optimal weighting improves performance on rare poses and improves the generalization of our method to different datasets. Our paper demonstrates the unique advantage of random forest representations: minimax estimation is straightforward with no significant retraining burden.",
keywords = "Human pose estimation, Minimax, Regression, Regression forests",
author = "{Daphne Tsatsoulis}, P. and David Forsyth",
year = "2015",
month = "1",
day = "1",
doi = "10.1007/978-3-319-16178-5_47",
language = "English (US)",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
publisher = "Springer-Verlag",
pages = "669--684",
editor = "Bronstein, {Michael M.} and Carsten Rother and Lourdes Agapito",
booktitle = "Computer Vision - ECCV 2014 Workshops, Proceedings",

}

TY - GEN

T1 - Easy minimax estimation with random forests for human pose estimation

AU - Daphne Tsatsoulis, P.

AU - Forsyth, David

PY - 2015/1/1

Y1 - 2015/1/1

N2 - We describe a method for human parsing that is straightforward and competes with state-of-the-art performance on standard datasets. Unlike the state-of-the-art, our method does not search for individual body parts or poselets. Instead, a regression forest is used to predict a body configuration in body-space. The output of this regression forest is then combined in a novel way. Instead of averaging the output of each tree in the forest we use minimax to calculate optimal weights for the trees. This optimal weighting improves performance on rare poses and improves the generalization of our method to different datasets. Our paper demonstrates the unique advantage of random forest representations: minimax estimation is straightforward with no significant retraining burden.

AB - We describe a method for human parsing that is straightforward and competes with state-of-the-art performance on standard datasets. Unlike the state-of-the-art, our method does not search for individual body parts or poselets. Instead, a regression forest is used to predict a body configuration in body-space. The output of this regression forest is then combined in a novel way. Instead of averaging the output of each tree in the forest we use minimax to calculate optimal weights for the trees. This optimal weighting improves performance on rare poses and improves the generalization of our method to different datasets. Our paper demonstrates the unique advantage of random forest representations: minimax estimation is straightforward with no significant retraining burden.

KW - Human pose estimation

KW - Minimax

KW - Regression

KW - Regression forests

UR - http://www.scopus.com/inward/record.url?scp=84925310886&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84925310886&partnerID=8YFLogxK

U2 - 10.1007/978-3-319-16178-5_47

DO - 10.1007/978-3-319-16178-5_47

M3 - Conference contribution

AN - SCOPUS:84925310886

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 669

EP - 684

BT - Computer Vision - ECCV 2014 Workshops, Proceedings

A2 - Bronstein, Michael M.

A2 - Rother, Carsten

A2 - Agapito, Lourdes

PB - Springer-Verlag

ER -