Human Activity Recognition Using 3D Orthogonally-projected EfficientNet on Radar Time-Range-Doppler Signature

Zeyu Wang, Chenglin Yao, Jianfeng Ren, Meili Feng, Xudong Jiang

Research output: Chapter in Book/Conference proceedingConference contributionpeer-review

Abstract

In radar activity recognition, 2D signal representations such as spectrogram, cepstrum and cadence velocity diagram are often utilized, while range information is often neglected. In this work, we propose to utilize the 3D time-range-Doppler (TRD) representation, and design a 3D Orthogonally-Projected EfficientNet (3D-OPEN) to effectively capture the discriminant information embedded in the 3D TRD cubes for accurate classification. The proposed model aggregates the discriminant information from three orthogonal planes projected from the 3D feature space. It alleviates the difficulty of 3D CNNs in exploiting sparse semantic abstractions directly from the high-dimensional 3D representation. The proposed method is evaluated on the Millimeter-Wave Radar Walking Dataset. It significantly and consistently outperforms the state-of-the-art methods for radar activity recognition.

Original languageEnglish
Title of host publication2022 2nd IEEE International Conference on Software Engineering and Artificial Intelligence, SEAI 2022
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages26-30
Number of pages5
ISBN (Electronic)9781665482233
DOIs
Publication statusPublished - 2022
Event2nd IEEE International Conference on Software Engineering and Artificial Intelligence, SEAI 2022 - Xiamen, China
Duration: 10 Jun 202212 Jun 2022

Publication series

Name2022 2nd IEEE International Conference on Software Engineering and Artificial Intelligence, SEAI 2022

Conference

Conference2nd IEEE International Conference on Software Engineering and Artificial Intelligence, SEAI 2022
Country/TerritoryChina
CityXiamen
Period10/06/2212/06/22

Keywords

  • 3D Orthogonally-projected EfficientNet
  • FMCW radar
  • human activity recognition
  • time-range-doppler representation

ASJC Scopus subject areas

  • Artificial Intelligence
  • Computer Science Applications
  • Computer Vision and Pattern Recognition
  • Software
  • Control and Optimization

Fingerprint

Dive into the research topics of 'Human Activity Recognition Using 3D Orthogonally-projected EfficientNet on Radar Time-Range-Doppler Signature'. Together they form a unique fingerprint.

Cite this