Single Run Action Detector over Video Stream - A Privacy Preserving Approach

Anbumalar Saravanan; Justin Sanchez; Hassan Ghasemzadeh; Aurelia Macabasco-O’Connell; Hamed Tabkhi

doi:10.1007/978-981-16-0575-8_7

Single Run Action Detector over Video Stream - A Privacy Preserving Approach

Anbumalar Saravanan, Justin Sanchez, Hassan Ghasemzadeh, Aurelia Macabasco-O’Connell, Hamed Tabkhi

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

1 Scopus citations

Abstract

This paper takes initial strides at designing and evaluating a vision-based system for privacy ensured activity monitoring. The proposed technology utilizing Artificial Intelligence (AI)-empowered proactive systems offering continuous monitoring, behavioral analysis, and modeling of human activities. To this end, this paper presents Single Run Action Detector (S-RAD) which is a real-time privacy-preserving action detector that performs end-to-end action localization and classification. It is based on Faster-RCNN combined with temporal shift modeling and segment based sampling to capture the human actions. Results on UCF-Sports and UR Fall dataset present comparable accuracy to State-of-the-Art approaches with significantly lower model size and computation demand and the ability for real-time execution on edge embedded device (e.g. Nvidia Jetson Xavier).

Original language	English (US)
Title of host publication	Deep Learning for Human Activity Recognition - 2nd International Workshop, DL-HAR 2020, Held in Conjunction with IJCAI-PRICAI 2020, Proceedings
Editors	Xiaoli Li, Min Wu, Zhenghua Chen, Le Zhang
Publisher	Springer Science and Business Media Deutschland GmbH
Pages	85-98
Number of pages	14
ISBN (Print)	9789811605741
DOIs	https://doi.org/10.1007/978-981-16-0575-8_7
State	Published - 2021
Externally published	Yes
Event	2nd International Workshop on Deep Learning for Human Activity Recognition, DL-HAR 2020, held in conjunction with IJCAI-PRICAI 2020 - Virtual, Online Duration: Jan 8 2021 → Jan 8 2021

Publication series

Name	Communications in Computer and Information Science
Volume	1370
ISSN (Print)	1865-0929
ISSN (Electronic)	1865-0937

Conference

Conference	2nd International Workshop on Deep Learning for Human Activity Recognition, DL-HAR 2020, held in conjunction with IJCAI-PRICAI 2020
City	Virtual, Online
Period	1/8/21 → 1/8/21

Keywords

Action detection
Deep learning
Edge computing
Real time
Spatial-temporal neural network

ASJC Scopus subject areas

General Computer Science
General Mathematics

Access to Document

10.1007/978-981-16-0575-8_7

Cite this

Saravanan, A., Sanchez, J., Ghasemzadeh, H., Macabasco-O’Connell, A., & Tabkhi, H. (2021). Single Run Action Detector over Video Stream - A Privacy Preserving Approach. In X. Li, M. Wu, Z. Chen, & L. Zhang (Eds.), Deep Learning for Human Activity Recognition - 2nd International Workshop, DL-HAR 2020, Held in Conjunction with IJCAI-PRICAI 2020, Proceedings (pp. 85-98). (Communications in Computer and Information Science; Vol. 1370). Springer Science and Business Media Deutschland GmbH. https://doi.org/10.1007/978-981-16-0575-8_7

Single Run Action Detector over Video Stream - A Privacy Preserving Approach. / Saravanan, Anbumalar; Sanchez, Justin; Ghasemzadeh, Hassan et al.
Deep Learning for Human Activity Recognition - 2nd International Workshop, DL-HAR 2020, Held in Conjunction with IJCAI-PRICAI 2020, Proceedings. ed. / Xiaoli Li; Min Wu; Zhenghua Chen; Le Zhang. Springer Science and Business Media Deutschland GmbH, 2021. p. 85-98 (Communications in Computer and Information Science; Vol. 1370).

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

Saravanan, A, Sanchez, J, Ghasemzadeh, H, Macabasco-O’Connell, A & Tabkhi, H 2021, Single Run Action Detector over Video Stream - A Privacy Preserving Approach. in X Li, M Wu, Z Chen & L Zhang (eds), Deep Learning for Human Activity Recognition - 2nd International Workshop, DL-HAR 2020, Held in Conjunction with IJCAI-PRICAI 2020, Proceedings. Communications in Computer and Information Science, vol. 1370, Springer Science and Business Media Deutschland GmbH, pp. 85-98, 2nd International Workshop on Deep Learning for Human Activity Recognition, DL-HAR 2020, held in conjunction with IJCAI-PRICAI 2020, Virtual, Online, 1/8/21. https://doi.org/10.1007/978-981-16-0575-8_7

Saravanan A, Sanchez J, Ghasemzadeh H, Macabasco-O’Connell A, Tabkhi H. Single Run Action Detector over Video Stream - A Privacy Preserving Approach. In Li X, Wu M, Chen Z, Zhang L, editors, Deep Learning for Human Activity Recognition - 2nd International Workshop, DL-HAR 2020, Held in Conjunction with IJCAI-PRICAI 2020, Proceedings. Springer Science and Business Media Deutschland GmbH. 2021. p. 85-98. (Communications in Computer and Information Science). doi: 10.1007/978-981-16-0575-8_7

Saravanan, Anbumalar ; Sanchez, Justin ; Ghasemzadeh, Hassan et al. / Single Run Action Detector over Video Stream - A Privacy Preserving Approach. Deep Learning for Human Activity Recognition - 2nd International Workshop, DL-HAR 2020, Held in Conjunction with IJCAI-PRICAI 2020, Proceedings. editor / Xiaoli Li ; Min Wu ; Zhenghua Chen ; Le Zhang. Springer Science and Business Media Deutschland GmbH, 2021. pp. 85-98 (Communications in Computer and Information Science).

@inproceedings{6c5d3e5eaffb4efbbc857475d792e751,

title = "Single Run Action Detector over Video Stream - A Privacy Preserving Approach",

abstract = "This paper takes initial strides at designing and evaluating a vision-based system for privacy ensured activity monitoring. The proposed technology utilizing Artificial Intelligence (AI)-empowered proactive systems offering continuous monitoring, behavioral analysis, and modeling of human activities. To this end, this paper presents Single Run Action Detector (S-RAD) which is a real-time privacy-preserving action detector that performs end-to-end action localization and classification. It is based on Faster-RCNN combined with temporal shift modeling and segment based sampling to capture the human actions. Results on UCF-Sports and UR Fall dataset present comparable accuracy to State-of-the-Art approaches with significantly lower model size and computation demand and the ability for real-time execution on edge embedded device (e.g. Nvidia Jetson Xavier).",

keywords = "Action detection, Deep learning, Edge computing, Real time, Spatial-temporal neural network",

author = "Anbumalar Saravanan and Justin Sanchez and Hassan Ghasemzadeh and Aurelia Macabasco-O{\textquoteright}Connell and Hamed Tabkhi",

note = "Publisher Copyright: {\textcopyright} 2021, Springer Nature Singapore Pte Ltd.; 2nd International Workshop on Deep Learning for Human Activity Recognition, DL-HAR 2020, held in conjunction with IJCAI-PRICAI 2020 ; Conference date: 08-01-2021 Through 08-01-2021",

year = "2021",

doi = "10.1007/978-981-16-0575-8_7",

language = "English (US)",

isbn = "9789811605741",

series = "Communications in Computer and Information Science",

publisher = "Springer Science and Business Media Deutschland GmbH",

pages = "85--98",

editor = "Xiaoli Li and Min Wu and Zhenghua Chen and Le Zhang",

booktitle = "Deep Learning for Human Activity Recognition - 2nd International Workshop, DL-HAR 2020, Held in Conjunction with IJCAI-PRICAI 2020, Proceedings",

address = "Germany",

}

TY - GEN

T1 - Single Run Action Detector over Video Stream - A Privacy Preserving Approach

AU - Saravanan, Anbumalar

AU - Sanchez, Justin

AU - Ghasemzadeh, Hassan

AU - Macabasco-O’Connell, Aurelia

AU - Tabkhi, Hamed

PY - 2021

Y1 - 2021

N2 - This paper takes initial strides at designing and evaluating a vision-based system for privacy ensured activity monitoring. The proposed technology utilizing Artificial Intelligence (AI)-empowered proactive systems offering continuous monitoring, behavioral analysis, and modeling of human activities. To this end, this paper presents Single Run Action Detector (S-RAD) which is a real-time privacy-preserving action detector that performs end-to-end action localization and classification. It is based on Faster-RCNN combined with temporal shift modeling and segment based sampling to capture the human actions. Results on UCF-Sports and UR Fall dataset present comparable accuracy to State-of-the-Art approaches with significantly lower model size and computation demand and the ability for real-time execution on edge embedded device (e.g. Nvidia Jetson Xavier).

AB - This paper takes initial strides at designing and evaluating a vision-based system for privacy ensured activity monitoring. The proposed technology utilizing Artificial Intelligence (AI)-empowered proactive systems offering continuous monitoring, behavioral analysis, and modeling of human activities. To this end, this paper presents Single Run Action Detector (S-RAD) which is a real-time privacy-preserving action detector that performs end-to-end action localization and classification. It is based on Faster-RCNN combined with temporal shift modeling and segment based sampling to capture the human actions. Results on UCF-Sports and UR Fall dataset present comparable accuracy to State-of-the-Art approaches with significantly lower model size and computation demand and the ability for real-time execution on edge embedded device (e.g. Nvidia Jetson Xavier).

KW - Action detection

KW - Deep learning

KW - Edge computing

KW - Real time

KW - Spatial-temporal neural network

UR - http://www.scopus.com/inward/record.url?scp=85102741517&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85102741517&partnerID=8YFLogxK

U2 - 10.1007/978-981-16-0575-8_7

DO - 10.1007/978-981-16-0575-8_7

M3 - Conference contribution

AN - SCOPUS:85102741517

SN - 9789811605741

T3 - Communications in Computer and Information Science

SP - 85

EP - 98

BT - Deep Learning for Human Activity Recognition - 2nd International Workshop, DL-HAR 2020, Held in Conjunction with IJCAI-PRICAI 2020, Proceedings

A2 - Li, Xiaoli

A2 - Wu, Min

A2 - Chen, Zhenghua

A2 - Zhang, Le

PB - Springer Science and Business Media Deutschland GmbH

T2 - 2nd International Workshop on Deep Learning for Human Activity Recognition, DL-HAR 2020, held in conjunction with IJCAI-PRICAI 2020

Y2 - 8 January 2021 through 8 January 2021

ER -

Single Run Action Detector over Video Stream - A Privacy Preserving Approach

Abstract

Publication series

Conference

Keywords

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this