Evaluation Metrics for Continuous Human Activity Classification Using Distributed Radar Networks

Ronny G. Guendel; Francesco  Fioranelli; Alexander Yarovoy

doi:10.1109/RadarConf2248738.2022.9764181

Evaluation Metrics for Continuous Human Activity Classification Using Distributed Radar Networks

Ronny G. Guendel, Francesco Fioranelli, Alexander Yarovoy

Microwave Sensing, Signals & Systems

Research output: Chapter in Book/Conference proceedings/Edited volume › Conference contribution › Scientific › peer-review

2 Citations (Scopus)

34 Downloads (Pure)

Abstract

Continuous Human Activity Recognition (HAR) in arbitrary directions is investigated using 5 spatially distributed pulsed Ultra-Wideband (UWB) radars. Such activities performed in arbitrary and unconstrained trajectories render a more natural occurrence of Activities of Daily Living (ADL) to be recognized. An innovative signal level fusion method was applied on the Range-Time (RT) maps, and deep learning classification via Recurrent Neural Networks (RNN) with and without bidi-rectionality was used on the computed micro-Doppler (μD) spectrogram. To assess classification performances, novel evaluation metrics accounting for the continuous nature of the sequence of activities and for imbalances in the dataset are proposed and compared with existing metrics. It is shown that conventional accuracy evaluation is too coarse, and that the proposed metrics need to be considered for a more comprehensive evaluation.

Original language	English
Title of host publication	2022 IEEE Radar Conference (RadarConf22) Proceedings
Place of Publication	Piscataway
Publisher	IEEE
Number of pages	6
ISBN (Electronic)	978-1-7281-5368-1
ISBN (Print)	978-1-7281-5369-8
DOIs	https://doi.org/10.1109/RadarConf2248738.2022.9764181
Publication status	Published - 2022
Event	2022 IEEE Radar Conference - New York City, United States Duration: 21 Mar 2022 → 25 Mar 2022

Publication series

Name	Proceedings of the IEEE Radar Conference
ISSN (Print)	1097-5764

Conference

Conference	2022 IEEE Radar Conference
Abbreviated title	RadarConf22
Country/Territory	United States
City	New York City
Period	21/03/22 → 25/03/22

Bibliographical note

Green Open Access added to TU Delft Institutional Repository 'You share, we take care!' - Taverne project https://www.openaccess.nl/en/you-share-we-take-care

Otherwise as indicated in the copyright section: the publisher is the copyright holder of this work and the author uses the Dutch legislation to make this work public.

Keywords

Micro-Doppler Classification
Distributed Radar
LSTM
Human Activity Recognition

Access to Document

10.1109/RadarConf2248738.2022.9764181

Evaluation_Metrics_for_Continuous_Human_Activity_Classification_Using_Distributed_Radar_NetworksFinal published version, 1.66 MB

Cite this

@inproceedings{53eb27074243464bbda93f9f218bbe84,

title = "Evaluation Metrics for Continuous Human Activity Classification Using Distributed Radar Networks",

abstract = "Continuous Human Activity Recognition (HAR) in arbitrary directions is investigated using 5 spatially distributed pulsed Ultra-Wideband (UWB) radars. Such activities performed in arbitrary and unconstrained trajectories render a more natural occurrence of Activities of Daily Living (ADL) to be recognized. An innovative signal level fusion method was applied on the Range-Time (RT) maps, and deep learning classification via Recurrent Neural Networks (RNN) with and without bidi-rectionality was used on the computed micro-Doppler (μD) spectrogram. To assess classification performances, novel evaluation metrics accounting for the continuous nature of the sequence of activities and for imbalances in the dataset are proposed and compared with existing metrics. It is shown that conventional accuracy evaluation is too coarse, and that the proposed metrics need to be considered for a more comprehensive evaluation.",

keywords = "Micro-Doppler Classification, Distributed Radar, LSTM, Human Activity Recognition",

author = "Guendel, {Ronny G.} and Francesco Fioranelli and Alexander Yarovoy",

note = "Green Open Access added to TU Delft Institutional Repository 'You share, we take care!' - Taverne project https://www.openaccess.nl/en/you-share-we-take-care Otherwise as indicated in the copyright section: the publisher is the copyright holder of this work and the author uses the Dutch legislation to make this work public.; 2022 IEEE Radar Conference <br/>, RadarConf22 ; Conference date: 21-03-2022 Through 25-03-2022",

year = "2022",

doi = "10.1109/RadarConf2248738.2022.9764181",

language = "English",

isbn = "978-1-7281-5369-8",

series = "Proceedings of the IEEE Radar Conference",

publisher = "IEEE",

booktitle = "2022 IEEE Radar Conference (RadarConf22) Proceedings",

address = "United States",

}

Guendel, RG , Fioranelli, F & Yarovoy, A 2022, Evaluation Metrics for Continuous Human Activity Classification Using Distributed Radar Networks. in 2022 IEEE Radar Conference (RadarConf22) Proceedings., 9764181, Proceedings of the IEEE Radar Conference, IEEE, Piscataway, 2022 IEEE Radar Conference
, New York City, United States, 21/03/22. https://doi.org/10.1109/RadarConf2248738.2022.9764181

Evaluation Metrics for Continuous Human Activity Classification Using Distributed Radar Networks. / Guendel, Ronny G.; Fioranelli, Francesco ; Yarovoy, Alexander.
2022 IEEE Radar Conference (RadarConf22) Proceedings. Piscataway: IEEE, 2022. 9764181 (Proceedings of the IEEE Radar Conference).

Research output: Chapter in Book/Conference proceedings/Edited volume › Conference contribution › Scientific › peer-review

TY - GEN

T1 - Evaluation Metrics for Continuous Human Activity Classification Using Distributed Radar Networks

AU - Guendel, Ronny G.

AU - Fioranelli, Francesco

AU - Yarovoy, Alexander

N1 - Green Open Access added to TU Delft Institutional Repository 'You share, we take care!' - Taverne project https://www.openaccess.nl/en/you-share-we-take-care Otherwise as indicated in the copyright section: the publisher is the copyright holder of this work and the author uses the Dutch legislation to make this work public.

PY - 2022

Y1 - 2022

N2 - Continuous Human Activity Recognition (HAR) in arbitrary directions is investigated using 5 spatially distributed pulsed Ultra-Wideband (UWB) radars. Such activities performed in arbitrary and unconstrained trajectories render a more natural occurrence of Activities of Daily Living (ADL) to be recognized. An innovative signal level fusion method was applied on the Range-Time (RT) maps, and deep learning classification via Recurrent Neural Networks (RNN) with and without bidi-rectionality was used on the computed micro-Doppler (μD) spectrogram. To assess classification performances, novel evaluation metrics accounting for the continuous nature of the sequence of activities and for imbalances in the dataset are proposed and compared with existing metrics. It is shown that conventional accuracy evaluation is too coarse, and that the proposed metrics need to be considered for a more comprehensive evaluation.

AB - Continuous Human Activity Recognition (HAR) in arbitrary directions is investigated using 5 spatially distributed pulsed Ultra-Wideband (UWB) radars. Such activities performed in arbitrary and unconstrained trajectories render a more natural occurrence of Activities of Daily Living (ADL) to be recognized. An innovative signal level fusion method was applied on the Range-Time (RT) maps, and deep learning classification via Recurrent Neural Networks (RNN) with and without bidi-rectionality was used on the computed micro-Doppler (μD) spectrogram. To assess classification performances, novel evaluation metrics accounting for the continuous nature of the sequence of activities and for imbalances in the dataset are proposed and compared with existing metrics. It is shown that conventional accuracy evaluation is too coarse, and that the proposed metrics need to be considered for a more comprehensive evaluation.

KW - Micro-Doppler Classification

KW - Distributed Radar

KW - LSTM

KW - Human Activity Recognition

UR - http://www.scopus.com/inward/record.url?scp=85128100573&partnerID=8YFLogxK

U2 - 10.1109/RadarConf2248738.2022.9764181

DO - 10.1109/RadarConf2248738.2022.9764181

M3 - Conference contribution

SN - 978-1-7281-5369-8

T3 - Proceedings of the IEEE Radar Conference

BT - 2022 IEEE Radar Conference (RadarConf22) Proceedings

PB - IEEE

CY - Piscataway

T2 - 2022 IEEE Radar Conference <br/>

Y2 - 21 March 2022 through 25 March 2022

ER -

Evaluation Metrics for Continuous Human Activity Classification Using Distributed Radar Networks

Abstract

Publication series

Conference

Bibliographical note

Keywords

Access to Document

Other files and links

Fingerprint

Cite this