Human action recognition based on sub-data learning

Yang Chen; Tian Wang; Jiakun Li; Xiaowei Lv; Hichem Snoussi

doi:10.1007/978-981-10-7305-2_52

Human action recognition based on sub-data learning

Yang Chen
, Tian Wang^*
, Jiakun Li
, Xiaowei Lv
, Hichem Snoussi

^*Corresponding author for this work

School of Artificial Intelligence(Institute of Artificial Intelligence)

Beihang University
China Electronics Technology Group Corporation
Université de technologie de Troyes

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

1 Scopus citations

Abstract

Human action recognizing nowadays plays a key role in varieties of computer vision applications while at the same time it’s quite challenging for the requirement of accuracy and robustness. Most current computer vision methods focus on algorithms designing classifiers with handcrafted features which are complex and inflexible. To automatically extract both spatial and temporal features, in this paper we propose a method of human action recognition based on sub-data learning which combines the proposed 3D convolutional neural network (3DCNN) with the One-versus-One (OvO) algorithm. We also employ effective data augmentation to reduce overfitting. We evaluate our method on the KTH and UCF Sports dataset and achieve promising results.

Original language	English
Title of host publication	Computer Vision - 2nd CCF Chinese Conference, CCCV 2017, Proceedings
Editors	Jinfeng Yang, Qingshan Liu, Liang Wang, Xiang Bai, Qinghua Hu, Ming-Ming Cheng, Deyu Meng
Publisher	Springer Verlag
Pages	617-626
Number of pages	10
ISBN (Print)	9789811073045
DOIs	https://doi.org/10.1007/978-981-10-7305-2_52
State	Published - 2017
Event	2nd Chinese Conference on Computer Vision, CCCV 2017 - Tianjin, China Duration: 11 Oct 2017 → 14 Oct 2017

Publication series

Name	Communications in Computer and Information Science
Volume	773
ISSN (Print)	1865-0929

Conference

Conference	2nd Chinese Conference on Computer Vision, CCCV 2017
Country/Territory	China
City	Tianjin
Period	11/10/17 → 14/10/17

Keywords

3DCNN
Action recognition
Sub-data learning

Access to Document

10.1007/978-981-10-7305-2_52

Cite this

Chen, Y., Wang, T., Li, J., Lv, X., & Snoussi, H. (2017). Human action recognition based on sub-data learning. In J. Yang, Q. Liu, L. Wang, X. Bai, Q. Hu, M.-M. Cheng, & D. Meng (Eds.), Computer Vision - 2nd CCF Chinese Conference, CCCV 2017, Proceedings (pp. 617-626). (Communications in Computer and Information Science; Vol. 773). Springer Verlag. https://doi.org/10.1007/978-981-10-7305-2_52

@inproceedings{d6f20b7aa05940029809bfbbff85632b,

title = "Human action recognition based on sub-data learning",

abstract = "Human action recognizing nowadays plays a key role in varieties of computer vision applications while at the same time it{\textquoteright}s quite challenging for the requirement of accuracy and robustness. Most current computer vision methods focus on algorithms designing classifiers with handcrafted features which are complex and inflexible. To automatically extract both spatial and temporal features, in this paper we propose a method of human action recognition based on sub-data learning which combines the proposed 3D convolutional neural network (3DCNN) with the One-versus-One (OvO) algorithm. We also employ effective data augmentation to reduce overfitting. We evaluate our method on the KTH and UCF Sports dataset and achieve promising results.",

keywords = "3DCNN, Action recognition, Sub-data learning",

author = "Yang Chen and Tian Wang and Jiakun Li and Xiaowei Lv and Hichem Snoussi",

note = "Publisher Copyright: {\textcopyright} 2017, Springer Nature Singapore Pte Ltd.; 2nd Chinese Conference on Computer Vision, CCCV 2017 ; Conference date: 11-10-2017 Through 14-10-2017",

year = "2017",

doi = "10.1007/978-981-10-7305-2\_52",

language = "英语",

isbn = "9789811073045",

series = "Communications in Computer and Information Science",

publisher = "Springer Verlag",

pages = "617--626",

editor = "Jinfeng Yang and Qingshan Liu and Liang Wang and Xiang Bai and Qinghua Hu and Ming-Ming Cheng and Deyu Meng",

booktitle = "Computer Vision - 2nd CCF Chinese Conference, CCCV 2017, Proceedings",

address = "德国",

}

Chen, Y, Wang, T, Li, J, Lv, X & Snoussi, H 2017, Human action recognition based on sub-data learning. in J Yang, Q Liu, L Wang, X Bai, Q Hu, M-M Cheng & D Meng (eds), Computer Vision - 2nd CCF Chinese Conference, CCCV 2017, Proceedings. Communications in Computer and Information Science, vol. 773, Springer Verlag, pp. 617-626, 2nd Chinese Conference on Computer Vision, CCCV 2017, Tianjin, China, 11/10/17. https://doi.org/10.1007/978-981-10-7305-2_52

Human action recognition based on sub-data learning. / Chen, Yang; Wang, Tian; Li, Jiakun et al.
Computer Vision - 2nd CCF Chinese Conference, CCCV 2017, Proceedings. ed. / Jinfeng Yang; Qingshan Liu; Liang Wang; Xiang Bai; Qinghua Hu; Ming-Ming Cheng; Deyu Meng. Springer Verlag, 2017. p. 617-626 (Communications in Computer and Information Science; Vol. 773).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - Human action recognition based on sub-data learning

AU - Chen, Yang

AU - Wang, Tian

AU - Li, Jiakun

AU - Lv, Xiaowei

AU - Snoussi, Hichem

PY - 2017

Y1 - 2017

N2 - Human action recognizing nowadays plays a key role in varieties of computer vision applications while at the same time it’s quite challenging for the requirement of accuracy and robustness. Most current computer vision methods focus on algorithms designing classifiers with handcrafted features which are complex and inflexible. To automatically extract both spatial and temporal features, in this paper we propose a method of human action recognition based on sub-data learning which combines the proposed 3D convolutional neural network (3DCNN) with the One-versus-One (OvO) algorithm. We also employ effective data augmentation to reduce overfitting. We evaluate our method on the KTH and UCF Sports dataset and achieve promising results.

AB - Human action recognizing nowadays plays a key role in varieties of computer vision applications while at the same time it’s quite challenging for the requirement of accuracy and robustness. Most current computer vision methods focus on algorithms designing classifiers with handcrafted features which are complex and inflexible. To automatically extract both spatial and temporal features, in this paper we propose a method of human action recognition based on sub-data learning which combines the proposed 3D convolutional neural network (3DCNN) with the One-versus-One (OvO) algorithm. We also employ effective data augmentation to reduce overfitting. We evaluate our method on the KTH and UCF Sports dataset and achieve promising results.

KW - 3DCNN

KW - Action recognition

KW - Sub-data learning

UR - https://www.scopus.com/pages/publications/85038125070

U2 - 10.1007/978-981-10-7305-2_52

DO - 10.1007/978-981-10-7305-2_52

M3 - 会议稿件

AN - SCOPUS:85038125070

SN - 9789811073045

T3 - Communications in Computer and Information Science

SP - 617

EP - 626

BT - Computer Vision - 2nd CCF Chinese Conference, CCCV 2017, Proceedings

A2 - Yang, Jinfeng

A2 - Liu, Qingshan

A2 - Wang, Liang

A2 - Bai, Xiang

A2 - Hu, Qinghua

A2 - Cheng, Ming-Ming

A2 - Meng, Deyu

PB - Springer Verlag

T2 - 2nd Chinese Conference on Computer Vision, CCCV 2017

Y2 - 11 October 2017 through 14 October 2017

ER -

Human action recognition based on sub-data learning

Abstract

Publication series

Conference

Keywords

Access to Document

Other files and links

Fingerprint

Cite this