Point Transformer-Based Human Activity Recognition Using High-Dimensional Radar Point Clouds

Zhongyuan  Guo; Ronny G. Guendel; Alexander Yarovoy; Francesco Fioranelli

doi:10.1109/RadarConf2351548.2023.10149679

Point Transformer-Based Human Activity Recognition Using High-Dimensional Radar Point Clouds

Zhongyuan Guo, Ronny G. Guendel, Alexander Yarovoy, Francesco Fioranelli

Microwave Sensing, Signals & Systems

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

1 Citation (Scopus)

11 Downloads (Pure)

Abstract

Radar-based Human Activity Recognition(HAR) is considered by using snapshots of point clouds. Such point cloudsinterpret 2D images generated by an mm-wave FMCW MIMO radar enriched byincluding Doppler and temporal information. We use the similarity between suchradar data representation and the core of the self-attention concept inartificial intelligence. Three self-attention models (Point Transformer) areinvestigated to classify Activities of Daily Living (ADL). An experimentaldataset collected at TU Delft is used to explore the best combination ofdifferent input features, the effect of a proposed Adaptive ClutterCancellation (ACC) method, and the robustness in a leave-one-subject-outscenario. Results with a macro F1 score in the order of 90% are demonstratedwith the proposed method, including activities that are static postures withlittle associated Doppler.

Original language	English
Title of host publication	Proceedings of the 2023 IEEE Radar Conference (RadarConf23)
Place of Publication	Piscataway
Publisher	IEEE
Pages	1-6
Number of pages	6
ISBN (Electronic)	978-1-6654-3669-4
ISBN (Print)	978-1-6654-3670-0
DOIs	https://doi.org/10.1109/RadarConf2351548.2023.10149679
Publication status	Published - 2023
Event	2023 IEEE Radar Conference (RadarConf23) - San Antonio, United States Duration: 1 May 2023 → 5 May 2023

Conference

Conference	2023 IEEE Radar Conference (RadarConf23)
Country/Territory	United States
City	San Antonio
Period	1/05/23 → 5/05/23

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

Human Activity Recognition
Imaging Radar
Deep Learning
Point Transformer
Activities of Daily Living

Access to Document

10.1109/RadarConf2351548.2023.10149679

Point_Transformer-Based_Human_Activity_Recognition_Using_High-Dimensional_Radar_Point_CloudsFinal published version, 1.52 MB

Cite this

@inproceedings{8b3f2a070af34089a6cb82a4201c848c,

title = "Point Transformer-Based Human Activity Recognition Using High-Dimensional Radar Point Clouds",

abstract = "Radar-based Human Activity Recognition(HAR) is considered by using snapshots of point clouds. Such point cloudsinterpret 2D images generated by an mm-wave FMCW MIMO radar enriched byincluding Doppler and temporal information. We use the similarity between suchradar data representation and the core of the self-attention concept inartificial intelligence. Three self-attention models (Point Transformer) areinvestigated to classify Activities of Daily Living (ADL). An experimentaldataset collected at TU Delft is used to explore the best combination ofdifferent input features, the effect of a proposed Adaptive ClutterCancellation (ACC) method, and the robustness in a leave-one-subject-outscenario. Results with a macro F1 score in the order of 90% are demonstratedwith the proposed method, including activities that are static postures withlittle associated Doppler.",

keywords = "Human Activity Recognition, Imaging Radar, Deep Learning, Point Transformer, Activities of Daily Living",

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

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.; 2023 IEEE Radar Conference (RadarConf23) ; Conference date: 01-05-2023 Through 05-05-2023",

year = "2023",

doi = "10.1109/RadarConf2351548.2023.10149679",

language = "English",

isbn = "978-1-6654-3670-0",

pages = "1--6",

booktitle = "Proceedings of the 2023 IEEE Radar Conference (RadarConf23)",

publisher = "IEEE",

address = "United States",

}

Guo, Z, Guendel, RG , Yarovoy, A & Fioranelli, F 2023, Point Transformer-Based Human Activity Recognition Using High-Dimensional Radar Point Clouds. in Proceedings of the 2023 IEEE Radar Conference (RadarConf23). IEEE, Piscataway, pp. 1-6, 2023 IEEE Radar Conference (RadarConf23), San Antonio, United States, 1/05/23. https://doi.org/10.1109/RadarConf2351548.2023.10149679

Point Transformer-Based Human Activity Recognition Using High-Dimensional Radar Point Clouds. / Guo, Zhongyuan ; Guendel, Ronny G.; Yarovoy, Alexander et al.
Proceedings of the 2023 IEEE Radar Conference (RadarConf23). Piscataway: IEEE, 2023. p. 1-6.

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

TY - GEN

T1 - Point Transformer-Based Human Activity Recognition Using High-Dimensional Radar Point Clouds

AU - Guo, Zhongyuan

AU - Guendel, Ronny G.

AU - Yarovoy, Alexander

AU - Fioranelli, Francesco

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 - 2023

Y1 - 2023

N2 - Radar-based Human Activity Recognition(HAR) is considered by using snapshots of point clouds. Such point cloudsinterpret 2D images generated by an mm-wave FMCW MIMO radar enriched byincluding Doppler and temporal information. We use the similarity between suchradar data representation and the core of the self-attention concept inartificial intelligence. Three self-attention models (Point Transformer) areinvestigated to classify Activities of Daily Living (ADL). An experimentaldataset collected at TU Delft is used to explore the best combination ofdifferent input features, the effect of a proposed Adaptive ClutterCancellation (ACC) method, and the robustness in a leave-one-subject-outscenario. Results with a macro F1 score in the order of 90% are demonstratedwith the proposed method, including activities that are static postures withlittle associated Doppler.

AB - Radar-based Human Activity Recognition(HAR) is considered by using snapshots of point clouds. Such point cloudsinterpret 2D images generated by an mm-wave FMCW MIMO radar enriched byincluding Doppler and temporal information. We use the similarity between suchradar data representation and the core of the self-attention concept inartificial intelligence. Three self-attention models (Point Transformer) areinvestigated to classify Activities of Daily Living (ADL). An experimentaldataset collected at TU Delft is used to explore the best combination ofdifferent input features, the effect of a proposed Adaptive ClutterCancellation (ACC) method, and the robustness in a leave-one-subject-outscenario. Results with a macro F1 score in the order of 90% are demonstratedwith the proposed method, including activities that are static postures withlittle associated Doppler.

KW - Human Activity Recognition

KW - Imaging Radar

KW - Deep Learning

KW - Point Transformer

KW - Activities of Daily Living

UR - http://www.scopus.com/inward/record.url?scp=85163710696&partnerID=8YFLogxK

U2 - 10.1109/RadarConf2351548.2023.10149679

DO - 10.1109/RadarConf2351548.2023.10149679

M3 - Conference contribution

SN - 978-1-6654-3670-0

SP - 1

EP - 6

BT - Proceedings of the 2023 IEEE Radar Conference (RadarConf23)

PB - IEEE

CY - Piscataway

T2 - 2023 IEEE Radar Conference (RadarConf23)

Y2 - 1 May 2023 through 5 May 2023

ER -

Point Transformer-Based Human Activity Recognition Using High-Dimensional Radar Point Clouds

Abstract

Conference

Bibliographical note

Keywords

Access to Document

Other files and links

Fingerprint

Cite this