Human Action Recognition from Depth Videos Using Pool of Multiple Projections with Greedy Selection

Chien-Quang LE; Sang PHAN; Thanh Duc NGO; Duy-Dinh LE; Shin'ichi SATOH; Duc Anh DUONG

doi:10.1587/transinf.2015EDP7430

Human Action Recognition from Depth Videos Using Pool of Multiple Projections with Greedy Selection

Chien-Quang LE, Sang PHAN, Thanh Duc NGO, Duy-Dinh LE, Shin'ichi SATOH, Duc Anh DUONG

Full Text Views

0

Share
Cite this

Summary :

Depth-based action recognition has been attracting the attention of researchers because of the advantages of depth cameras over standard RGB cameras. One of these advantages is that depth data can provide richer information from multiple projections. In particular, multiple projections can be used to extract discriminative motion patterns that would not be discernible from one fixed projection. However, high computational costs have meant that recent studies have exploited only a small number of projections, such as front, side, and top. Thus, a large number of projections, which may be useful for discriminating actions, are discarded. In this paper, we propose an efficient method to exploit pools of multiple projections for recognizing actions in depth videos. First, we project 3D data onto multiple 2D-planes from different viewpoints sampled on a geodesic dome to obtain a large number of projections. Then, we train and test action classifiers independently for each projection. To reduce the computational cost, we propose a greedy method to select a small yet robust combination of projections. The idea is that best complementary projections will be considered first when searching for optimal combination. We conducted extensive experiments to verify the effectiveness of our method on three challenging benchmarks: MSR Action 3D, MSR Gesture 3D, and 3D Action Pairs. The experimental results show that our method outperforms other state-of-the-art methods while using a small number of projections.

Publication: IEICE TRANSACTIONS on Information Vol.E99-D No.8 pp.2161-2171

Publication Date: 2016/08/01

Publicized: 2016/04/25

Online ISSN: 1745-1361

DOI: 10.1587/transinf.2015EDP7430

Type of Manuscript: PAPER

Category: Pattern Recognition

Authors

Chien-Quang LE
  Graduate University for Advanced Studies (SOKENDAI)
Sang PHAN
  National Institute of Informatics (NII)
Thanh Duc NGO
  University of Information Technology
Duy-Dinh LE
  National Institute of Informatics (NII)
Shin'ichi SATOH
  National Institute of Informatics (NII)
Duc Anh DUONG
  University of Information Technology

Keyword

action recognition, depth sequences, multiple projections, greedy method

Cite this

Copy

Chien-Quang LE, Sang PHAN, Thanh Duc NGO, Duy-Dinh LE, Shin'ichi SATOH, Duc Anh DUONG, "Human Action Recognition from Depth Videos Using Pool of Multiple Projections with Greedy Selection" in IEICE TRANSACTIONS on Information, vol. E99-D, no. 8, pp. 2161-2171, August 2016, doi: 10.1587/transinf.2015EDP7430.
Abstract: Depth-based action recognition has been attracting the attention of researchers because of the advantages of depth cameras over standard RGB cameras. One of these advantages is that depth data can provide richer information from multiple projections. In particular, multiple projections can be used to extract discriminative motion patterns that would not be discernible from one fixed projection. However, high computational costs have meant that recent studies have exploited only a small number of projections, such as front, side, and top. Thus, a large number of projections, which may be useful for discriminating actions, are discarded. In this paper, we propose an efficient method to exploit pools of multiple projections for recognizing actions in depth videos. First, we project 3D data onto multiple 2D-planes from different viewpoints sampled on a geodesic dome to obtain a large number of projections. Then, we train and test action classifiers independently for each projection. To reduce the computational cost, we propose a greedy method to select a small yet robust combination of projections. The idea is that best complementary projections will be considered first when searching for optimal combination. We conducted extensive experiments to verify the effectiveness of our method on three challenging benchmarks: MSR Action 3D, MSR Gesture 3D, and 3D Action Pairs. The experimental results show that our method outperforms other state-of-the-art methods while using a small number of projections.
URL: https://globals.ieice.org/en_transactions/information/10.1587/transinf.2015EDP7430/_p

Copy

@ARTICLE{e99-d_8_2161,
author={Chien-Quang LE, Sang PHAN, Thanh Duc NGO, Duy-Dinh LE, Shin'ichi SATOH, Duc Anh DUONG, },
journal={IEICE TRANSACTIONS on Information},
title={Human Action Recognition from Depth Videos Using Pool of Multiple Projections with Greedy Selection},
year={2016},
volume={E99-D},
number={8},
pages={2161-2171},
abstract={Depth-based action recognition has been attracting the attention of researchers because of the advantages of depth cameras over standard RGB cameras. One of these advantages is that depth data can provide richer information from multiple projections. In particular, multiple projections can be used to extract discriminative motion patterns that would not be discernible from one fixed projection. However, high computational costs have meant that recent studies have exploited only a small number of projections, such as front, side, and top. Thus, a large number of projections, which may be useful for discriminating actions, are discarded. In this paper, we propose an efficient method to exploit pools of multiple projections for recognizing actions in depth videos. First, we project 3D data onto multiple 2D-planes from different viewpoints sampled on a geodesic dome to obtain a large number of projections. Then, we train and test action classifiers independently for each projection. To reduce the computational cost, we propose a greedy method to select a small yet robust combination of projections. The idea is that best complementary projections will be considered first when searching for optimal combination. We conducted extensive experiments to verify the effectiveness of our method on three challenging benchmarks: MSR Action 3D, MSR Gesture 3D, and 3D Action Pairs. The experimental results show that our method outperforms other state-of-the-art methods while using a small number of projections.},
keywords={},
doi={10.1587/transinf.2015EDP7430},
ISSN={1745-1361},
month={August},}

Copy

TY - JOUR
TI - Human Action Recognition from Depth Videos Using Pool of Multiple Projections with Greedy Selection
T2 - IEICE TRANSACTIONS on Information
SP - 2161
EP - 2171
AU - Chien-Quang LE
AU - Sang PHAN
AU - Thanh Duc NGO
AU - Duy-Dinh LE
AU - Shin'ichi SATOH
AU - Duc Anh DUONG
PY - 2016
DO - 10.1587/transinf.2015EDP7430
JO - IEICE TRANSACTIONS on Information
SN - 1745-1361
VL - E99-D
IS - 8
JA - IEICE TRANSACTIONS on Information
Y1 - August 2016
AB - Depth-based action recognition has been attracting the attention of researchers because of the advantages of depth cameras over standard RGB cameras. One of these advantages is that depth data can provide richer information from multiple projections. In particular, multiple projections can be used to extract discriminative motion patterns that would not be discernible from one fixed projection. However, high computational costs have meant that recent studies have exploited only a small number of projections, such as front, side, and top. Thus, a large number of projections, which may be useful for discriminating actions, are discarded. In this paper, we propose an efficient method to exploit pools of multiple projections for recognizing actions in depth videos. First, we project 3D data onto multiple 2D-planes from different viewpoints sampled on a geodesic dome to obtain a large number of projections. Then, we train and test action classifiers independently for each projection. To reduce the computational cost, we propose a greedy method to select a small yet robust combination of projections. The idea is that best complementary projections will be considered first when searching for optimal combination. We conducted extensive experiments to verify the effectiveness of our method on three challenging benchmarks: MSR Action 3D, MSR Gesture 3D, and 3D Action Pairs. The experimental results show that our method outperforms other state-of-the-art methods while using a small number of projections.
ER -