RadarSleepNet: Sleep Pose Classification via PointNet++ and 5D Radar Point Clouds

Alessandra Fusco; Mervenur Akkus; Nastassia Vysotskaya; Souvik Hazra; Lorenzo Servadei; Andreas Maier; Robert Wille

doi:10.1109/MAPCON58678.2023.10463831

RadarSleepNet: Sleep Pose Classification via PointNet++ and 5D Radar Point Clouds

Alessandra Fusco
, Mervenur Akkus
, Nastassia Vysotskaya
, Souvik Hazra
, Lorenzo Servadei
, Andreas Maier
, Robert Wille

Chair of Design Automation

Infineon Technologies AG
Technical University of Munich
Friedrich-Alexander Universitat Erlangen-Nurnberg (FAU)

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

7 Scopus citations

Abstract

Understanding sleep patterns and postures is critical for assessing overall well-being. However, traditional sleep analysis methods are often limited in their practicality due to invasive devices or complex configurations. In this study, we introduce RadarSleepNet, a non-intrusive 60 GHz Frequency-modulated Continuous Wave (FMCW) radar-based system for sleep posture monitoring that accurately infers sleep postures without compromising privacy or comfort, even in low-light conditions. Our system combines a SincNet classifier and a PointNet++ sleep pose estimation model, achieving remarkable class accuracy for each sleep posture: 98.43% for supine, 98.01% for side (chest facing radar), 97.22% for prone, and 95.72% for side (back facing radar). This demonstrates its effectiveness in accurately classifying sleep postures. This innovation offers significant potential in healthcare, providing insights into disease management and improving individual health understanding.

Original language	English
Title of host publication	2023 IEEE Microwaves, Antennas, and Propagation Conference, MAPCON 2023
Publisher	Institute of Electrical and Electronics Engineers Inc.
ISBN (Electronic)	9798350328264
DOIs	https://doi.org/10.1109/MAPCON58678.2023.10463831
State	Published - 2023
Event	2023 IEEE Microwaves, Antennas, and Propagation Conference, MAPCON 2023 - Ahmedabad, India Duration: 11 Dec 2023 → 14 Dec 2023

Publication series

Name	2023 IEEE Microwaves, Antennas, and Propagation Conference, MAPCON 2023

Conference

Conference	2023 IEEE Microwaves, Antennas, and Propagation Conference, MAPCON 2023
Country/Territory	India
City	Ahmedabad
Period	11/12/23 → 14/12/23

Keywords

FMCW radar
PointNet++
SincNet
contactless
deep learning
sleep monitoring

Access to Document

10.1109/MAPCON58678.2023.10463831

Cite this

Fusco, A., Akkus, M., Vysotskaya, N., Hazra, S., Servadei, L., Maier, A., & Wille, R. (2023). RadarSleepNet: Sleep Pose Classification via PointNet++ and 5D Radar Point Clouds. In 2023 IEEE Microwaves, Antennas, and Propagation Conference, MAPCON 2023 (2023 IEEE Microwaves, Antennas, and Propagation Conference, MAPCON 2023). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/MAPCON58678.2023.10463831

@inproceedings{e2547e7c1ef74ab7beac2ee340edb0fd,

title = "RadarSleepNet: Sleep Pose Classification via PointNet++ and 5D Radar Point Clouds",

abstract = "Understanding sleep patterns and postures is critical for assessing overall well-being. However, traditional sleep analysis methods are often limited in their practicality due to invasive devices or complex configurations. In this study, we introduce RadarSleepNet, a non-intrusive 60 GHz Frequency-modulated Continuous Wave (FMCW) radar-based system for sleep posture monitoring that accurately infers sleep postures without compromising privacy or comfort, even in low-light conditions. Our system combines a SincNet classifier and a PointNet++ sleep pose estimation model, achieving remarkable class accuracy for each sleep posture: 98.43\% for supine, 98.01\% for side (chest facing radar), 97.22\% for prone, and 95.72\% for side (back facing radar). This demonstrates its effectiveness in accurately classifying sleep postures. This innovation offers significant potential in healthcare, providing insights into disease management and improving individual health understanding.",

keywords = "FMCW radar, PointNet++, SincNet, contactless, deep learning, sleep monitoring",

author = "Alessandra Fusco and Mervenur Akkus and Nastassia Vysotskaya and Souvik Hazra and Lorenzo Servadei and Andreas Maier and Robert Wille",

note = "Publisher Copyright: {\textcopyright} 2023 IEEE.; 2023 IEEE Microwaves, Antennas, and Propagation Conference, MAPCON 2023 ; Conference date: 11-12-2023 Through 14-12-2023",

year = "2023",

doi = "10.1109/MAPCON58678.2023.10463831",

language = "English",

series = "2023 IEEE Microwaves, Antennas, and Propagation Conference, MAPCON 2023",

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

booktitle = "2023 IEEE Microwaves, Antennas, and Propagation Conference, MAPCON 2023",

}

Fusco, A, Akkus, M, Vysotskaya, N, Hazra, S, Servadei, L, Maier, A & Wille, R 2023, RadarSleepNet: Sleep Pose Classification via PointNet++ and 5D Radar Point Clouds. in 2023 IEEE Microwaves, Antennas, and Propagation Conference, MAPCON 2023. 2023 IEEE Microwaves, Antennas, and Propagation Conference, MAPCON 2023, Institute of Electrical and Electronics Engineers Inc., 2023 IEEE Microwaves, Antennas, and Propagation Conference, MAPCON 2023, Ahmedabad, India, 11/12/23. https://doi.org/10.1109/MAPCON58678.2023.10463831

RadarSleepNet: Sleep Pose Classification via PointNet++ and 5D Radar Point Clouds. / Fusco, Alessandra; Akkus, Mervenur; Vysotskaya, Nastassia et al.
2023 IEEE Microwaves, Antennas, and Propagation Conference, MAPCON 2023. Institute of Electrical and Electronics Engineers Inc., 2023. (2023 IEEE Microwaves, Antennas, and Propagation Conference, MAPCON 2023).

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

TY - GEN

T1 - RadarSleepNet

T2 - 2023 IEEE Microwaves, Antennas, and Propagation Conference, MAPCON 2023

AU - Fusco, Alessandra

AU - Akkus, Mervenur

AU - Vysotskaya, Nastassia

AU - Hazra, Souvik

AU - Servadei, Lorenzo

AU - Maier, Andreas

AU - Wille, Robert

PY - 2023

Y1 - 2023

N2 - Understanding sleep patterns and postures is critical for assessing overall well-being. However, traditional sleep analysis methods are often limited in their practicality due to invasive devices or complex configurations. In this study, we introduce RadarSleepNet, a non-intrusive 60 GHz Frequency-modulated Continuous Wave (FMCW) radar-based system for sleep posture monitoring that accurately infers sleep postures without compromising privacy or comfort, even in low-light conditions. Our system combines a SincNet classifier and a PointNet++ sleep pose estimation model, achieving remarkable class accuracy for each sleep posture: 98.43% for supine, 98.01% for side (chest facing radar), 97.22% for prone, and 95.72% for side (back facing radar). This demonstrates its effectiveness in accurately classifying sleep postures. This innovation offers significant potential in healthcare, providing insights into disease management and improving individual health understanding.

AB - Understanding sleep patterns and postures is critical for assessing overall well-being. However, traditional sleep analysis methods are often limited in their practicality due to invasive devices or complex configurations. In this study, we introduce RadarSleepNet, a non-intrusive 60 GHz Frequency-modulated Continuous Wave (FMCW) radar-based system for sleep posture monitoring that accurately infers sleep postures without compromising privacy or comfort, even in low-light conditions. Our system combines a SincNet classifier and a PointNet++ sleep pose estimation model, achieving remarkable class accuracy for each sleep posture: 98.43% for supine, 98.01% for side (chest facing radar), 97.22% for prone, and 95.72% for side (back facing radar). This demonstrates its effectiveness in accurately classifying sleep postures. This innovation offers significant potential in healthcare, providing insights into disease management and improving individual health understanding.

KW - FMCW radar

KW - PointNet++

KW - SincNet

KW - contactless

KW - deep learning

KW - sleep monitoring

UR - https://www.scopus.com/pages/publications/85190365665

U2 - 10.1109/MAPCON58678.2023.10463831

DO - 10.1109/MAPCON58678.2023.10463831

M3 - Conference contribution

AN - SCOPUS:85190365665

T3 - 2023 IEEE Microwaves, Antennas, and Propagation Conference, MAPCON 2023

BT - 2023 IEEE Microwaves, Antennas, and Propagation Conference, MAPCON 2023

PB - Institute of Electrical and Electronics Engineers Inc.

Y2 - 11 December 2023 through 14 December 2023

ER -

Fusco A, Akkus M, Vysotskaya N, Hazra S, Servadei L, Maier A et al. RadarSleepNet: Sleep Pose Classification via PointNet++ and 5D Radar Point Clouds. In 2023 IEEE Microwaves, Antennas, and Propagation Conference, MAPCON 2023. Institute of Electrical and Electronics Engineers Inc. 2023. (2023 IEEE Microwaves, Antennas, and Propagation Conference, MAPCON 2023). doi: 10.1109/MAPCON58678.2023.10463831

RadarSleepNet: Sleep Pose Classification via PointNet++ and 5D Radar Point Clouds

Abstract

Publication series

Conference

Keywords

Access to Document

Other files and links

Fingerprint

Cite this