Human motion classification by micro-doppler radar using intelligent algorithms.

Saved in:

Bibliographic Details
Title:	Human motion classification by micro-doppler radar using intelligent algorithms.
Authors:	Ballen, Andres Felipe Arias¹ juan.martinezq@unimilitar.edu.co, Cuesta, Edith Paola Estupiñán¹ edith.estupinan@unimilitar.edu.co, Quintero, Juan Carlos Martínez¹ est.andres.arias1@unimilitar.edu.co
Source:	International Journal of Electrical & Computer Engineering (2088-8708). Feb2025, Vol. 15 Issue 1, p455-466. 12p.
Subjects:	Continuous wave radar, Signal generators, Image recognition (Computer vision), Convolutional neural networks, Human mechanics, Motion, Radio frequency
Abstract:	This article introduces a technique for detecting four human movements using micro-doppler radar and intelligent algorithms. Micro-doppler radar exhibits the capability to detect and measure object movements with intricate detail, even capturing complex or non-rigid motions, while accurately identifying direction, velocity, and motion patterns. The application of intelligent algorithms enhances detection efficiency and reduces false alarms by discerning subtle movement patterns, thereby facilitating more accurate detection and a deeper understanding of observed object dynamics. A continuous wave radar setup was implemented utilizing a spectrum analyzer and radio frequency (RF) generator capturing signals in a spectrogram centered at 2,395 MHz. Six models were assessed for image classification: VGG-16, VGG-19, MobileNet, MobileNet V2, Xception, and Inception V3. A dataset comprising 500 images depicting four movements-running, walking, arm raising, and jumping-was curated. Our findings reveal that the most optimal architecture in terms of training time, accuracy, and loss is VGG-16, achieving an accuracy of 96%. Furthermore, precision values of 96%, 100%, and 98% were obtained for the movements of walking, running, and arm raising, respectively. Notably, VGG-16 exhibited a training loss of 4.191E-04, attributed to the utilization of the Adam optimizer with a learning rate of 0.001 over 15 epochs and a batch size of 32. [ABSTRACT FROM AUTHOR]
	Copyright of International Journal of Electrical & Computer Engineering (2088-8708) is the property of Institute of Advanced Engineering & Science and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)
Database:	Engineering Source

Description
Abstract:	This article introduces a technique for detecting four human movements using micro-doppler radar and intelligent algorithms. Micro-doppler radar exhibits the capability to detect and measure object movements with intricate detail, even capturing complex or non-rigid motions, while accurately identifying direction, velocity, and motion patterns. The application of intelligent algorithms enhances detection efficiency and reduces false alarms by discerning subtle movement patterns, thereby facilitating more accurate detection and a deeper understanding of observed object dynamics. A continuous wave radar setup was implemented utilizing a spectrum analyzer and radio frequency (RF) generator capturing signals in a spectrogram centered at 2,395 MHz. Six models were assessed for image classification: VGG-16, VGG-19, MobileNet, MobileNet V2, Xception, and Inception V3. A dataset comprising 500 images depicting four movements-running, walking, arm raising, and jumping-was curated. Our findings reveal that the most optimal architecture in terms of training time, accuracy, and loss is VGG-16, achieving an accuracy of 96%. Furthermore, precision values of 96%, 100%, and 98% were obtained for the movements of walking, running, and arm raising, respectively. Notably, VGG-16 exhibited a training loss of 4.191E-04, attributed to the utilization of the Adam optimizer with a learning rate of 0.001 over 15 epochs and a batch size of 32. [ABSTRACT FROM AUTHOR]
ISSN:	20888708
DOI:	10.11591/ijece.v15i1.pp455-466