Siamese Networks Based People Tracking for 360-degree Videos with Equi-angular Cubemap Format

Kuan Chen Tai; Chih Wei Tang

doi:10.1109/ICCE-Taiwan49838.2020.9258313

Siamese Networks Based People Tracking for 360-degree Videos with Equi-angular Cubemap Format

Kuan Chen Tai, Chih Wei Tang

Department of Communication Engineering

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

2 Scopus citations

Abstract

This paper proposes a deep learning based pedestrian tracking scheme for 360-degree videos using equiangular cubemap (EAC) format. To be robust against content discontinuity of EAC images, this paper proposes an efficient face stitching scheme such that the tracker keeps tracking across adjacent faces and avoids raising geometric deformation simultaneously. By referring to statistics of score maps from efficient fully-convolutional siamese networks, the proposed mechanism of template update determines the timing of update. Experimental results show that the proposed tracker operates at 60 fps and outperforms the fully convolutional siamese networks based tracker on 360-degree videos with EAC format both in precision plots and success plots.

Original language	English
Title of host publication	2020 IEEE International Conference on Consumer Electronics - Taiwan, ICCE-Taiwan 2020
Publisher	Institute of Electrical and Electronics Engineers Inc.
ISBN (Electronic)	9781728173993
DOIs	https://doi.org/10.1109/ICCE-Taiwan49838.2020.9258313
State	Published - 28 Sep 2020
Event	7th IEEE International Conference on Consumer Electronics - Taiwan, ICCE-Taiwan 2020 - Taoyuan, Taiwan Duration: 28 Sep 2020 → 30 Sep 2020

Publication series

Name	2020 IEEE International Conference on Consumer Electronics - Taiwan, ICCE-Taiwan 2020

Conference

Conference	7th IEEE International Conference on Consumer Electronics - Taiwan, ICCE-Taiwan 2020
Country/Territory	Taiwan
City	Taoyuan
Period	28/09/20 → 30/09/20

Access to Document

10.1109/ICCE-Taiwan49838.2020.9258313

Deep Learning Based 360-Degree Vi Deo Processing and Analysis
Tang, C.-W. (PI)
1/08/19 → 31/07/20
Project: Research
360-Degree Vi Deo Processing and Analysis for Vi Rtual Reality Applications
Tang, C.-W. (PI)
1/08/18 → 31/07/19
Project: Research

Cite this

Tai, K. C., & Tang, C. W. (2020). Siamese Networks Based People Tracking for 360-degree Videos with Equi-angular Cubemap Format. In 2020 IEEE International Conference on Consumer Electronics - Taiwan, ICCE-Taiwan 2020 Article 9258313 (2020 IEEE International Conference on Consumer Electronics - Taiwan, ICCE-Taiwan 2020). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/ICCE-Taiwan49838.2020.9258313

@inproceedings{aa0888e81ac54839851b63990bf83376,

title = "Siamese Networks Based People Tracking for 360-degree Videos with Equi-angular Cubemap Format",

abstract = "This paper proposes a deep learning based pedestrian tracking scheme for 360-degree videos using equiangular cubemap (EAC) format. To be robust against content discontinuity of EAC images, this paper proposes an efficient face stitching scheme such that the tracker keeps tracking across adjacent faces and avoids raising geometric deformation simultaneously. By referring to statistics of score maps from efficient fully-convolutional siamese networks, the proposed mechanism of template update determines the timing of update. Experimental results show that the proposed tracker operates at 60 fps and outperforms the fully convolutional siamese networks based tracker on 360-degree videos with EAC format both in precision plots and success plots.",

author = "Tai, {Kuan Chen} and Tang, {Chih Wei}",

note = "Publisher Copyright: {\textcopyright} 2020 IEEE.; 7th IEEE International Conference on Consumer Electronics - Taiwan, ICCE-Taiwan 2020 ; Conference date: 28-09-2020 Through 30-09-2020",

year = "2020",

month = sep,

day = "28",

doi = "10.1109/ICCE-Taiwan49838.2020.9258313",

language = "???core.languages.en_GB???",

series = "2020 IEEE International Conference on Consumer Electronics - Taiwan, ICCE-Taiwan 2020",

publisher = "Institute of Electrical and Electronics Engineers Inc.",

booktitle = "2020 IEEE International Conference on Consumer Electronics - Taiwan, ICCE-Taiwan 2020",

}

Tai, KC & Tang, CW 2020, Siamese Networks Based People Tracking for 360-degree Videos with Equi-angular Cubemap Format. in 2020 IEEE International Conference on Consumer Electronics - Taiwan, ICCE-Taiwan 2020., 9258313, 2020 IEEE International Conference on Consumer Electronics - Taiwan, ICCE-Taiwan 2020, Institute of Electrical and Electronics Engineers Inc., 7th IEEE International Conference on Consumer Electronics - Taiwan, ICCE-Taiwan 2020, Taoyuan, Taiwan, 28/09/20. https://doi.org/10.1109/ICCE-Taiwan49838.2020.9258313

Siamese Networks Based People Tracking for 360-degree Videos with Equi-angular Cubemap Format. / Tai, Kuan Chen; Tang, Chih Wei.
2020 IEEE International Conference on Consumer Electronics - Taiwan, ICCE-Taiwan 2020. Institute of Electrical and Electronics Engineers Inc., 2020. 9258313 (2020 IEEE International Conference on Consumer Electronics - Taiwan, ICCE-Taiwan 2020).

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

TY - GEN

T1 - Siamese Networks Based People Tracking for 360-degree Videos with Equi-angular Cubemap Format

AU - Tai, Kuan Chen

AU - Tang, Chih Wei

PY - 2020/9/28

Y1 - 2020/9/28

N2 - This paper proposes a deep learning based pedestrian tracking scheme for 360-degree videos using equiangular cubemap (EAC) format. To be robust against content discontinuity of EAC images, this paper proposes an efficient face stitching scheme such that the tracker keeps tracking across adjacent faces and avoids raising geometric deformation simultaneously. By referring to statistics of score maps from efficient fully-convolutional siamese networks, the proposed mechanism of template update determines the timing of update. Experimental results show that the proposed tracker operates at 60 fps and outperforms the fully convolutional siamese networks based tracker on 360-degree videos with EAC format both in precision plots and success plots.

AB - This paper proposes a deep learning based pedestrian tracking scheme for 360-degree videos using equiangular cubemap (EAC) format. To be robust against content discontinuity of EAC images, this paper proposes an efficient face stitching scheme such that the tracker keeps tracking across adjacent faces and avoids raising geometric deformation simultaneously. By referring to statistics of score maps from efficient fully-convolutional siamese networks, the proposed mechanism of template update determines the timing of update. Experimental results show that the proposed tracker operates at 60 fps and outperforms the fully convolutional siamese networks based tracker on 360-degree videos with EAC format both in precision plots and success plots.

UR - http://www.scopus.com/inward/record.url?scp=85098455752&partnerID=8YFLogxK

U2 - 10.1109/ICCE-Taiwan49838.2020.9258313

DO - 10.1109/ICCE-Taiwan49838.2020.9258313

M3 - 會議論文篇章

AN - SCOPUS:85098455752

T3 - 2020 IEEE International Conference on Consumer Electronics - Taiwan, ICCE-Taiwan 2020

BT - 2020 IEEE International Conference on Consumer Electronics - Taiwan, ICCE-Taiwan 2020

PB - Institute of Electrical and Electronics Engineers Inc.

T2 - 7th IEEE International Conference on Consumer Electronics - Taiwan, ICCE-Taiwan 2020

Y2 - 28 September 2020 through 30 September 2020

ER -

Tai KC, Tang CW. Siamese Networks Based People Tracking for 360-degree Videos with Equi-angular Cubemap Format. In 2020 IEEE International Conference on Consumer Electronics - Taiwan, ICCE-Taiwan 2020. Institute of Electrical and Electronics Engineers Inc. 2020. 9258313. (2020 IEEE International Conference on Consumer Electronics - Taiwan, ICCE-Taiwan 2020). doi: 10.1109/ICCE-Taiwan49838.2020.9258313

Siamese Networks Based People Tracking for 360-degree Videos with Equi-angular Cubemap Format

Abstract

Publication series

Conference

Access to Document

Fingerprint

Projects

Deep Learning Based 360-Degree Vi Deo Processing and Analysis

360-Degree Vi Deo Processing and Analysis for Vi Rtual Reality Applications

Cite this