3D Convolutional Neural network for Home Monitoring using Low Resolution Thermal-sensor Array

Lili Tao, Timothy Volonakis, Bo Tan, Ziqi Zhang, Yanguo Jing, Melvyn Smith

Research output: Contribution to conferencePaper

Abstract

The recognition of daily actions, such as walking, sitting or
standing, in the home is informative for assisted living, smart
homes and general health care. A variety of actions in complex
scenes can be recognised using visual information. However
cameras succumb to privacy concerns. In this paper, we present
a home activity recognition system using an 8×8 infared sensor array. This low spatial resolution retains user privacy, but is still a powerful representation of actions in a scene. Actions are recognised using a 3D convolutional neural network, extracting not only spatial but temporal information from video sequences. Experimental results obtained from a publicly available dataset Infra-ADL2018 demonstrate a better performance of the proposed approach compared to the state-of-the-art. We show that the sensor is considered better at detecting the occurrence of falls and activities of daily living. Our method achieves an overall accuracy of 97.22% across 7 actions with a fall detection sensitivity of 100% and specificity of 99.31%
Original languageEnglish
Publication statusAccepted/In press - 21 Feb 2019
Event3rd IET International Conference on Technologies for Active and Assisted Living - IET London: Savoy Place, London, United Kingdom
Duration: 25 Mar 201925 Mar 2019
https://events.theiet.org/techaal/

Conference

Conference3rd IET International Conference on Technologies for Active and Assisted Living
CountryUnited Kingdom
CityLondon
Period25/03/1925/03/19
Internet address

Fingerprint

Sensor arrays
Health care
Neural networks
Monitoring
Sensors
Hot Temperature
Assisted living

Cite this

Tao, L., Volonakis, T., Tan, B., Zhang, Z., Jing, Y., & Smith, M. (Accepted/In press). 3D Convolutional Neural network for Home Monitoring using Low Resolution Thermal-sensor Array. Paper presented at 3rd IET International Conference on Technologies for Active and Assisted Living, London, United Kingdom.

3D Convolutional Neural network for Home Monitoring using Low Resolution Thermal-sensor Array. / Tao, Lili; Volonakis, Timothy ; Tan, Bo; Zhang, Ziqi; Jing, Yanguo; Smith, Melvyn .

2019. Paper presented at 3rd IET International Conference on Technologies for Active and Assisted Living, London, United Kingdom.

Research output: Contribution to conferencePaper

Tao, L, Volonakis, T, Tan, B, Zhang, Z, Jing, Y & Smith, M 2019, '3D Convolutional Neural network for Home Monitoring using Low Resolution Thermal-sensor Array' Paper presented at 3rd IET International Conference on Technologies for Active and Assisted Living, London, United Kingdom, 25/03/19 - 25/03/19, .
Tao L, Volonakis T, Tan B, Zhang Z, Jing Y, Smith M. 3D Convolutional Neural network for Home Monitoring using Low Resolution Thermal-sensor Array. 2019. Paper presented at 3rd IET International Conference on Technologies for Active and Assisted Living, London, United Kingdom.
Tao, Lili ; Volonakis, Timothy ; Tan, Bo ; Zhang, Ziqi ; Jing, Yanguo ; Smith, Melvyn . / 3D Convolutional Neural network for Home Monitoring using Low Resolution Thermal-sensor Array. Paper presented at 3rd IET International Conference on Technologies for Active and Assisted Living, London, United Kingdom.
@conference{506a83278d054db9827afd9c15c4c716,
title = "3D Convolutional Neural network for Home Monitoring using Low Resolution Thermal-sensor Array",
abstract = "The recognition of daily actions, such as walking, sitting orstanding, in the home is informative for assisted living, smarthomes and general health care. A variety of actions in complexscenes can be recognised using visual information. Howevercameras succumb to privacy concerns. In this paper, we presenta home activity recognition system using an 8×8 infared sensor array. This low spatial resolution retains user privacy, but is still a powerful representation of actions in a scene. Actions are recognised using a 3D convolutional neural network, extracting not only spatial but temporal information from video sequences. Experimental results obtained from a publicly available dataset Infra-ADL2018 demonstrate a better performance of the proposed approach compared to the state-of-the-art. We show that the sensor is considered better at detecting the occurrence of falls and activities of daily living. Our method achieves an overall accuracy of 97.22{\%} across 7 actions with a fall detection sensitivity of 100{\%} and specificity of 99.31{\%}",
author = "Lili Tao and Timothy Volonakis and Bo Tan and Ziqi Zhang and Yanguo Jing and Melvyn Smith",
year = "2019",
month = "2",
day = "21",
language = "English",
note = "3rd IET International Conference on Technologies for Active and Assisted Living ; Conference date: 25-03-2019 Through 25-03-2019",
url = "https://events.theiet.org/techaal/",

}

TY - CONF

T1 - 3D Convolutional Neural network for Home Monitoring using Low Resolution Thermal-sensor Array

AU - Tao, Lili

AU - Volonakis, Timothy

AU - Tan, Bo

AU - Zhang, Ziqi

AU - Jing, Yanguo

AU - Smith, Melvyn

PY - 2019/2/21

Y1 - 2019/2/21

N2 - The recognition of daily actions, such as walking, sitting orstanding, in the home is informative for assisted living, smarthomes and general health care. A variety of actions in complexscenes can be recognised using visual information. Howevercameras succumb to privacy concerns. In this paper, we presenta home activity recognition system using an 8×8 infared sensor array. This low spatial resolution retains user privacy, but is still a powerful representation of actions in a scene. Actions are recognised using a 3D convolutional neural network, extracting not only spatial but temporal information from video sequences. Experimental results obtained from a publicly available dataset Infra-ADL2018 demonstrate a better performance of the proposed approach compared to the state-of-the-art. We show that the sensor is considered better at detecting the occurrence of falls and activities of daily living. Our method achieves an overall accuracy of 97.22% across 7 actions with a fall detection sensitivity of 100% and specificity of 99.31%

AB - The recognition of daily actions, such as walking, sitting orstanding, in the home is informative for assisted living, smarthomes and general health care. A variety of actions in complexscenes can be recognised using visual information. Howevercameras succumb to privacy concerns. In this paper, we presenta home activity recognition system using an 8×8 infared sensor array. This low spatial resolution retains user privacy, but is still a powerful representation of actions in a scene. Actions are recognised using a 3D convolutional neural network, extracting not only spatial but temporal information from video sequences. Experimental results obtained from a publicly available dataset Infra-ADL2018 demonstrate a better performance of the proposed approach compared to the state-of-the-art. We show that the sensor is considered better at detecting the occurrence of falls and activities of daily living. Our method achieves an overall accuracy of 97.22% across 7 actions with a fall detection sensitivity of 100% and specificity of 99.31%

M3 - Paper

ER -