Action Recognition Using Weighted Locality-Constrained Linear Coding

Jiangfeng YANG; Zheng MA

doi:10.1587/transinf.2014EDL8134

Action Recognition Using Weighted Locality-Constrained Linear Coding

Jiangfeng YANG, Zheng MA

Full Text Views

0

Share
Cite this

Summary :

Recently, locality-constrained linear coding (LLC) as a coding strategy has attracted much attention, due to its better reconstruction than sparse coding and vector quantization. However, LLC ignores the weight information of codewords during the coding stage, and assumes that every selected base has same credibility, even if their weights are different. To further improve the discriminative power of LLC code, we propose a weighted LLC algorithm that considers the codeword weight information. Experiments on the KTH and UCF datasets show that the recognition system based on WLLC achieves better performance than that based on the classical LLC and VQ, and outperforms the recent classical systems.

Publication: IEICE TRANSACTIONS on Information Vol.E98-D No.2 pp.462-466

Publication Date: 2015/02/01

Publicized: 2014/10/31

Online ISSN: 1745-1361

DOI: 10.1587/transinf.2014EDL8134

Type of Manuscript: LETTER

Category: Image Processing and Video Processing

Authors

Jiangfeng YANG
University of Electronic Science and Technology of China
Zheng MA
University of Electronic Science and Technology of China

Keyword

action recognition, action representation, weighted locality-constrained linear coding

Cite this

Copy

Jiangfeng YANG, Zheng MA, "Action Recognition Using Weighted Locality-Constrained Linear Coding" in IEICE TRANSACTIONS on Information, vol. E98-D, no. 2, pp. 462-466, February 2015, doi: 10.1587/transinf.2014EDL8134.
Abstract: Recently, locality-constrained linear coding (LLC) as a coding strategy has attracted much attention, due to its better reconstruction than sparse coding and vector quantization. However, LLC ignores the weight information of codewords during the coding stage, and assumes that every selected base has same credibility, even if their weights are different. To further improve the discriminative power of LLC code, we propose a weighted LLC algorithm that considers the codeword weight information. Experiments on the KTH and UCF datasets show that the recognition system based on WLLC achieves better performance than that based on the classical LLC and VQ, and outperforms the recent classical systems.
URL: https://globals.ieice.org/en_transactions/information/10.1587/transinf.2014EDL8134/_p

Copy

@ARTICLE{e98-d_2_462,
author={Jiangfeng YANG, Zheng MA, },
journal={IEICE TRANSACTIONS on Information},
title={Action Recognition Using Weighted Locality-Constrained Linear Coding},
year={2015},
volume={E98-D},
number={2},
pages={462-466},
abstract={Recently, locality-constrained linear coding (LLC) as a coding strategy has attracted much attention, due to its better reconstruction than sparse coding and vector quantization. However, LLC ignores the weight information of codewords during the coding stage, and assumes that every selected base has same credibility, even if their weights are different. To further improve the discriminative power of LLC code, we propose a weighted LLC algorithm that considers the codeword weight information. Experiments on the KTH and UCF datasets show that the recognition system based on WLLC achieves better performance than that based on the classical LLC and VQ, and outperforms the recent classical systems.},
keywords={},
doi={10.1587/transinf.2014EDL8134},
ISSN={1745-1361},
month={February},}

Copy

TY - JOUR
TI - Action Recognition Using Weighted Locality-Constrained Linear Coding
T2 - IEICE TRANSACTIONS on Information
SP - 462
EP - 466
AU - Jiangfeng YANG
AU - Zheng MA
PY - 2015
DO - 10.1587/transinf.2014EDL8134
JO - IEICE TRANSACTIONS on Information
SN - 1745-1361
VL - E98-D
IS - 2
JA - IEICE TRANSACTIONS on Information
Y1 - February 2015
AB - Recently, locality-constrained linear coding (LLC) as a coding strategy has attracted much attention, due to its better reconstruction than sparse coding and vector quantization. However, LLC ignores the weight information of codewords during the coding stage, and assumes that every selected base has same credibility, even if their weights are different. To further improve the discriminative power of LLC code, we propose a weighted LLC algorithm that considers the codeword weight information. Experiments on the KTH and UCF datasets show that the recognition system based on WLLC achieves better performance than that based on the classical LLC and VQ, and outperforms the recent classical systems.
ER -