Dop-DenseNet: Densely Convolutional Neural Network-Based Gesture Recognition Using a Micro-Doppler Radar
Hand gesture recognition is an efficient and practical solution for the non-contact human–machine interaction in smart devices. To date, vision-based methods are widely used in this research area, but they are susceptible to light conditions. To address this issue, radar-based gesture recognition us...
Main Authors: | , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
The Korean Institute of Electromagnetic Engineering and Science
2022-05-01
|
Series: | Journal of Electromagnetic Engineering and Science |
Subjects: | |
Online Access: | http://jees.kr/upload/pdf/jees-2022-3-r-95.pdf |
_version_ | 1811214638172340224 |
---|---|
author | Hai Le Van-Phuc Hoang Van Sang Doan Dai Phong Le |
author_facet | Hai Le Van-Phuc Hoang Van Sang Doan Dai Phong Le |
author_sort | Hai Le |
collection | DOAJ |
description | Hand gesture recognition is an efficient and practical solution for the non-contact human–machine interaction in smart devices. To date, vision-based methods are widely used in this research area, but they are susceptible to light conditions. To address this issue, radar-based gesture recognition using micro-Doppler signatures can be applied as an alternative. Accordingly, the use of a novel densely convolutional neural network model, Dop-DenseNet, is proposed in this paper for improving hand gesture recognition in terms of classification accuracy and latency. The model was designed with cross or skip connections in a dense architecture so that the former features, which can be lost in the forward-propagation process, can be reused. We evaluated our model with different numbers of filter channels and experimented with it using the Dop-Net dataset, with different time lengths of input data. As a result, it was found that the model with 64 3 × 3 filters and 200 time bins of micro-Doppler spectrogram data could achieve the best performance trade-off, with 99.87% classification accuracy and 3.1 ms latency. In comparison, our model remarkably outperformed the selected state-of-the-art neural networks (GoogLeNet, Res-Net-50, NasNet-Mobile, and MobileNet-V2) using the same Dop-Net dataset. |
first_indexed | 2024-04-12T06:08:08Z |
format | Article |
id | doaj.art-46814bcdc46b48218d015cfbc642b62a |
institution | Directory Open Access Journal |
issn | 2671-7255 2671-7263 |
language | English |
last_indexed | 2024-04-12T06:08:08Z |
publishDate | 2022-05-01 |
publisher | The Korean Institute of Electromagnetic Engineering and Science |
record_format | Article |
series | Journal of Electromagnetic Engineering and Science |
spelling | doaj.art-46814bcdc46b48218d015cfbc642b62a2022-12-22T03:44:47ZengThe Korean Institute of Electromagnetic Engineering and ScienceJournal of Electromagnetic Engineering and Science2671-72552671-72632022-05-0122333534310.26866/jees.2022.3.r.953515Dop-DenseNet: Densely Convolutional Neural Network-Based Gesture Recognition Using a Micro-Doppler RadarHai Le0Van-Phuc Hoang1Van Sang Doan2Dai Phong Le3 Institute of System Integration, Le Quy Don Technical University, Hanoi, Vietnam Institute of System Integration, Le Quy Don Technical University, Hanoi, Vietnam Vietnam Naval Academy, Nha Trang, Vietnam Institute of System Integration, Le Quy Don Technical University, Hanoi, VietnamHand gesture recognition is an efficient and practical solution for the non-contact human–machine interaction in smart devices. To date, vision-based methods are widely used in this research area, but they are susceptible to light conditions. To address this issue, radar-based gesture recognition using micro-Doppler signatures can be applied as an alternative. Accordingly, the use of a novel densely convolutional neural network model, Dop-DenseNet, is proposed in this paper for improving hand gesture recognition in terms of classification accuracy and latency. The model was designed with cross or skip connections in a dense architecture so that the former features, which can be lost in the forward-propagation process, can be reused. We evaluated our model with different numbers of filter channels and experimented with it using the Dop-Net dataset, with different time lengths of input data. As a result, it was found that the model with 64 3 × 3 filters and 200 time bins of micro-Doppler spectrogram data could achieve the best performance trade-off, with 99.87% classification accuracy and 3.1 ms latency. In comparison, our model remarkably outperformed the selected state-of-the-art neural networks (GoogLeNet, Res-Net-50, NasNet-Mobile, and MobileNet-V2) using the same Dop-Net dataset.http://jees.kr/upload/pdf/jees-2022-3-r-95.pdfconvolutional neural networkhand gesture recognitionmicro-doppler radar |
spellingShingle | Hai Le Van-Phuc Hoang Van Sang Doan Dai Phong Le Dop-DenseNet: Densely Convolutional Neural Network-Based Gesture Recognition Using a Micro-Doppler Radar Journal of Electromagnetic Engineering and Science convolutional neural network hand gesture recognition micro-doppler radar |
title | Dop-DenseNet: Densely Convolutional Neural Network-Based Gesture Recognition Using a Micro-Doppler Radar |
title_full | Dop-DenseNet: Densely Convolutional Neural Network-Based Gesture Recognition Using a Micro-Doppler Radar |
title_fullStr | Dop-DenseNet: Densely Convolutional Neural Network-Based Gesture Recognition Using a Micro-Doppler Radar |
title_full_unstemmed | Dop-DenseNet: Densely Convolutional Neural Network-Based Gesture Recognition Using a Micro-Doppler Radar |
title_short | Dop-DenseNet: Densely Convolutional Neural Network-Based Gesture Recognition Using a Micro-Doppler Radar |
title_sort | dop densenet densely convolutional neural network based gesture recognition using a micro doppler radar |
topic | convolutional neural network hand gesture recognition micro-doppler radar |
url | http://jees.kr/upload/pdf/jees-2022-3-r-95.pdf |
work_keys_str_mv | AT haile dopdensenetdenselyconvolutionalneuralnetworkbasedgesturerecognitionusingamicrodopplerradar AT vanphuchoang dopdensenetdenselyconvolutionalneuralnetworkbasedgesturerecognitionusingamicrodopplerradar AT vansangdoan dopdensenetdenselyconvolutionalneuralnetworkbasedgesturerecognitionusingamicrodopplerradar AT daiphongle dopdensenetdenselyconvolutionalneuralnetworkbasedgesturerecognitionusingamicrodopplerradar |