Multi-scale bidirectional local template patterns for real-time human detection

Jiu Xu, Ning Jiang, Xinwei Xue, Heming Sun, Wenxin Yu, Satoshi Goto

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

Abstract

In this paper, a feature named multi-scale bidirectional local template patterns (MBLTP) is proposed for human detection. As an extension of bidirectional local template patterns (BLTP), MBLTP not only integrates the textural and gradient information according to the four predefined templates but also calculates information for additional feature vectors by adjusting the scale of the training samples. These additional feature vectors contain multi-scale information on the samples, which can make the feature more discriminative than its original form. Experimental results for an INRIA dataset show that the detection rate of our proposed MBLTP feature outperforms those of other features such as the multi-level histogram of orientated gradient (multi-level HOG), multi scale block histogram of template (MB-HOT), and HOG-LBP. Moreover, in order to make our feature meet real-time requirements, an implementation based on a graphic process unit (GPU) is adopted to accelerate the calculation.

Original languageEnglish
Title of host publication2013 IEEE International Workshop on Multimedia Signal Processing, MMSP 2013
Pages379-383
Number of pages5
DOIs
Publication statusPublished - 2013
Event2013 IEEE 15th International Workshop on Multimedia Signal Processing, MMSP 2013 - Pula, Sardinia
Duration: 2013 Sep 302013 Oct 2

Other

Other2013 IEEE 15th International Workshop on Multimedia Signal Processing, MMSP 2013
CityPula, Sardinia
Period13/9/3013/10/2

Fingerprint

Chemical reactions

ASJC Scopus subject areas

  • Signal Processing

Cite this

Xu, J., Jiang, N., Xue, X., Sun, H., Yu, W., & Goto, S. (2013). Multi-scale bidirectional local template patterns for real-time human detection. In 2013 IEEE International Workshop on Multimedia Signal Processing, MMSP 2013 (pp. 379-383). [6659318] https://doi.org/10.1109/MMSP.2013.6659318

Multi-scale bidirectional local template patterns for real-time human detection. / Xu, Jiu; Jiang, Ning; Xue, Xinwei; Sun, Heming; Yu, Wenxin; Goto, Satoshi.

2013 IEEE International Workshop on Multimedia Signal Processing, MMSP 2013. 2013. p. 379-383 6659318.

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

Xu, J, Jiang, N, Xue, X, Sun, H, Yu, W & Goto, S 2013, Multi-scale bidirectional local template patterns for real-time human detection. in 2013 IEEE International Workshop on Multimedia Signal Processing, MMSP 2013., 6659318, pp. 379-383, 2013 IEEE 15th International Workshop on Multimedia Signal Processing, MMSP 2013, Pula, Sardinia, 13/9/30. https://doi.org/10.1109/MMSP.2013.6659318
Xu J, Jiang N, Xue X, Sun H, Yu W, Goto S. Multi-scale bidirectional local template patterns for real-time human detection. In 2013 IEEE International Workshop on Multimedia Signal Processing, MMSP 2013. 2013. p. 379-383. 6659318 https://doi.org/10.1109/MMSP.2013.6659318
Xu, Jiu ; Jiang, Ning ; Xue, Xinwei ; Sun, Heming ; Yu, Wenxin ; Goto, Satoshi. / Multi-scale bidirectional local template patterns for real-time human detection. 2013 IEEE International Workshop on Multimedia Signal Processing, MMSP 2013. 2013. pp. 379-383
@inproceedings{218b5692118141218db356526107fcf5,
title = "Multi-scale bidirectional local template patterns for real-time human detection",
abstract = "In this paper, a feature named multi-scale bidirectional local template patterns (MBLTP) is proposed for human detection. As an extension of bidirectional local template patterns (BLTP), MBLTP not only integrates the textural and gradient information according to the four predefined templates but also calculates information for additional feature vectors by adjusting the scale of the training samples. These additional feature vectors contain multi-scale information on the samples, which can make the feature more discriminative than its original form. Experimental results for an INRIA dataset show that the detection rate of our proposed MBLTP feature outperforms those of other features such as the multi-level histogram of orientated gradient (multi-level HOG), multi scale block histogram of template (MB-HOT), and HOG-LBP. Moreover, in order to make our feature meet real-time requirements, an implementation based on a graphic process unit (GPU) is adopted to accelerate the calculation.",
author = "Jiu Xu and Ning Jiang and Xinwei Xue and Heming Sun and Wenxin Yu and Satoshi Goto",
year = "2013",
doi = "10.1109/MMSP.2013.6659318",
language = "English",
isbn = "9781479901258",
pages = "379--383",
booktitle = "2013 IEEE International Workshop on Multimedia Signal Processing, MMSP 2013",

}

TY - GEN

T1 - Multi-scale bidirectional local template patterns for real-time human detection

AU - Xu, Jiu

AU - Jiang, Ning

AU - Xue, Xinwei

AU - Sun, Heming

AU - Yu, Wenxin

AU - Goto, Satoshi

PY - 2013

Y1 - 2013

N2 - In this paper, a feature named multi-scale bidirectional local template patterns (MBLTP) is proposed for human detection. As an extension of bidirectional local template patterns (BLTP), MBLTP not only integrates the textural and gradient information according to the four predefined templates but also calculates information for additional feature vectors by adjusting the scale of the training samples. These additional feature vectors contain multi-scale information on the samples, which can make the feature more discriminative than its original form. Experimental results for an INRIA dataset show that the detection rate of our proposed MBLTP feature outperforms those of other features such as the multi-level histogram of orientated gradient (multi-level HOG), multi scale block histogram of template (MB-HOT), and HOG-LBP. Moreover, in order to make our feature meet real-time requirements, an implementation based on a graphic process unit (GPU) is adopted to accelerate the calculation.

AB - In this paper, a feature named multi-scale bidirectional local template patterns (MBLTP) is proposed for human detection. As an extension of bidirectional local template patterns (BLTP), MBLTP not only integrates the textural and gradient information according to the four predefined templates but also calculates information for additional feature vectors by adjusting the scale of the training samples. These additional feature vectors contain multi-scale information on the samples, which can make the feature more discriminative than its original form. Experimental results for an INRIA dataset show that the detection rate of our proposed MBLTP feature outperforms those of other features such as the multi-level histogram of orientated gradient (multi-level HOG), multi scale block histogram of template (MB-HOT), and HOG-LBP. Moreover, in order to make our feature meet real-time requirements, an implementation based on a graphic process unit (GPU) is adopted to accelerate the calculation.

UR - http://www.scopus.com/inward/record.url?scp=84892525008&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84892525008&partnerID=8YFLogxK

U2 - 10.1109/MMSP.2013.6659318

DO - 10.1109/MMSP.2013.6659318

M3 - Conference contribution

SN - 9781479901258

SP - 379

EP - 383

BT - 2013 IEEE International Workshop on Multimedia Signal Processing, MMSP 2013

ER -