Robust Face Alignment with Random Forest: Analysis of Initialization, Landmarks Regression, and Shape Regularization Methods

Chun Fui LIEW; Takehisa YAIRI

doi:10.1587/transinf.2015EDP7154

Robust Face Alignment with Random Forest: Analysis of Initialization, Landmarks Regression, and Shape Regularization Methods

Chun Fui LIEW, Takehisa YAIRI

Full Text Views

0

Share
Cite this

Summary :

Random forest regressor has recently been proposed as a local landmark estimator in the face alignment problem. It has been shown that random forest regressor can achieve accurate, fast, and robust performance when coupled with a global face-shape regularizer. In this paper, we extend this approach and propose a new Local Forest Classification and Regression (LFCR) framework in order to handle face images with large yaw angles. Specifically, the LFCR has an additional classification step prior to the regression step. Our experiment results show that this additional classification step is useful in rejecting outliers prior to the regression step, thus improving the face alignment results. We also analyze each system component through detailed experiments. In addition to the selection of feature descriptors and several important tuning parameters of the random forest regressor, we examine different initialization and shape regularization processes. We compare our best outcomes to the state-of-the-art system and show that our method outperforms other parametric shape-fitting approaches.

Publication: IEICE TRANSACTIONS on Information Vol.E99-D No.2 pp.496-504

Publication Date: 2016/02/01

Publicized: 2015/10/27

Online ISSN: 1745-1361

DOI: 10.1587/transinf.2015EDP7154

Type of Manuscript: PAPER

Category: Image Recognition, Computer Vision

Authors

Chun Fui LIEW
The University of Tokyo
Takehisa YAIRI
The University of Tokyo

Keyword

face alignment, shape model fitting, random forest, point distribution model, sparse representation model

Cite this

Copy

Chun Fui LIEW, Takehisa YAIRI, "Robust Face Alignment with Random Forest: Analysis of Initialization, Landmarks Regression, and Shape Regularization Methods" in IEICE TRANSACTIONS on Information, vol. E99-D, no. 2, pp. 496-504, February 2016, doi: 10.1587/transinf.2015EDP7154.
Abstract: Random forest regressor has recently been proposed as a local landmark estimator in the face alignment problem. It has been shown that random forest regressor can achieve accurate, fast, and robust performance when coupled with a global face-shape regularizer. In this paper, we extend this approach and propose a new Local Forest Classification and Regression (LFCR) framework in order to handle face images with large yaw angles. Specifically, the LFCR has an additional classification step prior to the regression step. Our experiment results show that this additional classification step is useful in rejecting outliers prior to the regression step, thus improving the face alignment results. We also analyze each system component through detailed experiments. In addition to the selection of feature descriptors and several important tuning parameters of the random forest regressor, we examine different initialization and shape regularization processes. We compare our best outcomes to the state-of-the-art system and show that our method outperforms other parametric shape-fitting approaches.
URL: https://globals.ieice.org/en_transactions/information/10.1587/transinf.2015EDP7154/_p

Copy

@ARTICLE{e99-d_2_496,
author={Chun Fui LIEW, Takehisa YAIRI, },
journal={IEICE TRANSACTIONS on Information},
title={Robust Face Alignment with Random Forest: Analysis of Initialization, Landmarks Regression, and Shape Regularization Methods},
year={2016},
volume={E99-D},
number={2},
pages={496-504},
abstract={Random forest regressor has recently been proposed as a local landmark estimator in the face alignment problem. It has been shown that random forest regressor can achieve accurate, fast, and robust performance when coupled with a global face-shape regularizer. In this paper, we extend this approach and propose a new Local Forest Classification and Regression (LFCR) framework in order to handle face images with large yaw angles. Specifically, the LFCR has an additional classification step prior to the regression step. Our experiment results show that this additional classification step is useful in rejecting outliers prior to the regression step, thus improving the face alignment results. We also analyze each system component through detailed experiments. In addition to the selection of feature descriptors and several important tuning parameters of the random forest regressor, we examine different initialization and shape regularization processes. We compare our best outcomes to the state-of-the-art system and show that our method outperforms other parametric shape-fitting approaches.},
keywords={},
doi={10.1587/transinf.2015EDP7154},
ISSN={1745-1361},
month={February},}

Copy

TY - JOUR
TI - Robust Face Alignment with Random Forest: Analysis of Initialization, Landmarks Regression, and Shape Regularization Methods
T2 - IEICE TRANSACTIONS on Information
SP - 496
EP - 504
AU - Chun Fui LIEW
AU - Takehisa YAIRI
PY - 2016
DO - 10.1587/transinf.2015EDP7154
JO - IEICE TRANSACTIONS on Information
SN - 1745-1361
VL - E99-D
IS - 2
JA - IEICE TRANSACTIONS on Information
Y1 - February 2016
AB - Random forest regressor has recently been proposed as a local landmark estimator in the face alignment problem. It has been shown that random forest regressor can achieve accurate, fast, and robust performance when coupled with a global face-shape regularizer. In this paper, we extend this approach and propose a new Local Forest Classification and Regression (LFCR) framework in order to handle face images with large yaw angles. Specifically, the LFCR has an additional classification step prior to the regression step. Our experiment results show that this additional classification step is useful in rejecting outliers prior to the regression step, thus improving the face alignment results. We also analyze each system component through detailed experiments. In addition to the selection of feature descriptors and several important tuning parameters of the random forest regressor, we examine different initialization and shape regularization processes. We compare our best outcomes to the state-of-the-art system and show that our method outperforms other parametric shape-fitting approaches.
ER -