Computer Vision Based Millimeter Wave Radar For Vital Signal Detection

The increasing demand for contact-free vital signal monitoring in healthcare has drawn the attention of many researchers to develop remote vital signal detection systems using radar technology. One type of millimeter wave radar, known as frequency modulated continuous wave (FMCW) radar, has recently shown promising results in detecting the heartbeat and respiration rates of a single participant. However, versatile multiperson contact-free vital signal detection still remains challenging. The focus of this thesis is to answer the question: can we use computer vision technol- ogy to enable simultaneous multiperson vital signal detection with FMCW millimeter wave radar? We propose a comprehensive radar system that uses a computer vision component to identify and localize human objects in the environment and applies the recognized objects’ relative positions to help FMCW radar target and extract multiple persons’ heartbeat and respiration rates simultaneously. To achieve this, we utilize the Mask Region-based Convo- lutional Neural Network (Mask R-CNN) computer vision model to accurately identify and locate human objects in the scene and integrate it with an off-the-shelf FMCW radar to collect raw waveform data for vital signal extraction. To validate the system, we conduct experiments with varying distances and angles involv- ing one person in the scene, as well as with two persons in the scene, one at a fixed distance and angle, and the other person at a fixed angle but various distances. We collect vital signals through wearable vital signal collection devices to verify the accuracy of the radar system’s vital signal detection results. Overall, in both one-person and two-person environments, our radar system yields similar vital signal estimation results to previous work’s one-person environment results, with an average of 2.63 beats per minute (bpm) heartbeat rate (HR) error and an average of 1.48 bpm respiration rate (RR) error. Thus, the proposed radar system shows great potential for detecting and extracting vital signals from multiple individuals in a scene simultaneously, which could prove to be an essential tool for healthcare professionals in diagnosing and treating patients.

Name thesis:degree_name

M.S.

Level thesis:degree_level

Thesis

Discipline thesis:degree_discipline

Computer Science

Grantor

University of Illinois at Urbana-Champaign

Year dc:date

2023

Author dc:creator

• Jin, Xin

Contributors dc:contributor

• Caesar, Matthew

• Fmcw Radar
• Vital Signal
• Computer Vision

Statement dc:rights

• Copyright 2023 Xin Jin

Language dc:language

en, eng

Handle dc:identifier

https://hdl.handle.net/2142/120209

OAI record oai:identifier

https://www.ideals.illinois.edu/oai-pmh?verb=GetRecord&metadataPrefix=etdms&identifier=oai%3Awww.ideals.illinois.edu%3A2142%2F120209