The original paper is in English. Non-English content has been machine-translated and may contain typographical errors or mistranslations. ex. Some numerals are expressed as "XNUMX".
Copyrights notice
The original paper is in English. Non-English content has been machine-translated and may contain typographical errors or mistranslations. Copyrights notice
Auparavant, les réseaux de neurones convolutifs ont fait d'énormes progrès dans la reconnaissance de cibles basée sur le radar micro-Doppler. Cependant, ces études n’ont pris en compte que la présence d’une seule cible à la fois dans la zone de surveillance. La reconnaissance simultanée de plusieurs cibles pour les radars de surveillance reste un problème assez difficile. Pour atténuer ce problème, cette lettre développe une stratégie d'apprentissage multi-instances multi-étiquettes (MIML), qui peut localiser automatiquement les modèles d'entrée cruciaux qui déclenchent les étiquettes. Bénéficiant de sa puissante capacité de découverte de relations cible-étiquette, le cadre proposé peut être formé avec une supervision limitée. Nous soulignons que seuls les échos de cibles uniques sont impliqués dans les données d'entraînement, évitant ainsi la préparation et l'annotation d'écho multi-cibles lors de la phase d'entraînement. Pour vérifier la validité de la méthode proposée, nous modélisons deux cibles terrestres représentatives, à savoir des personnes et des véhicules à roues, et effectuons de nombreuses expériences comparatives. Le résultat démontre que le cadre développé peut reconnaître simultanément plusieurs cibles et est également robuste à la variation du rapport signal/bruit (SNR), de la position initiale des cibles et de la différence de coefficient de diffusion.
Jingyi ZHANG
Nanjing Normal University
Kuiyu CHEN
Nanjing University of Science and Technology
Yue MA
Nanjing University of Science and Technology
The copyright of the original papers published on this site belongs to IEICE. Unauthorized use of the original or translated papers is prohibited. See IEICE Provisions on Copyright for details.
Copier
Jingyi ZHANG, Kuiyu CHEN, Yue MA, "Multi-Target Recognition Utilizing Micro-Doppler Signatures with Limited Supervision" in IEICE TRANSACTIONS on Electronics,
vol. E106-C, no. 8, pp. 454-457, August 2023, doi: 10.1587/transele.2022ECS6011.
Abstract: Previously, convolutional neural networks have made tremendous progress in target recognition based on micro-Doppler radar. However, these studies only considered the presence of one target at a time in the surveillance area. Simultaneous multi-targets recognition for surveillance radar remains a pretty challenging issue. To alleviate this issue, this letter develops a multi-instance multi-label (MIML) learning strategy, which can automatically locate the crucial input patterns that trigger the labels. Benefitting from its powerful target-label relation discovery ability, the proposed framework can be trained with limited supervision. We emphasize that only echoes from single targets are involved in training data, avoiding the preparation and annotation of multi-targets echo in the training stage. To verify the validity of the proposed method, we model two representative ground moving targets, i.e., person and wheeled vehicles, and carry out numerous comparative experiments. The result demonstrates that the developed framework can simultaneously recognize multiple targets and is also robust to variation of the signal-to-noise ratio (SNR), the initial position of targets, and the difference in scattering coefficient.
URL: https://global.ieice.org/en_transactions/electronics/10.1587/transele.2022ECS6011/_p
Copier
@ARTICLE{e106-c_8_454,
author={Jingyi ZHANG, Kuiyu CHEN, Yue MA, },
journal={IEICE TRANSACTIONS on Electronics},
title={Multi-Target Recognition Utilizing Micro-Doppler Signatures with Limited Supervision},
year={2023},
volume={E106-C},
number={8},
pages={454-457},
abstract={Previously, convolutional neural networks have made tremendous progress in target recognition based on micro-Doppler radar. However, these studies only considered the presence of one target at a time in the surveillance area. Simultaneous multi-targets recognition for surveillance radar remains a pretty challenging issue. To alleviate this issue, this letter develops a multi-instance multi-label (MIML) learning strategy, which can automatically locate the crucial input patterns that trigger the labels. Benefitting from its powerful target-label relation discovery ability, the proposed framework can be trained with limited supervision. We emphasize that only echoes from single targets are involved in training data, avoiding the preparation and annotation of multi-targets echo in the training stage. To verify the validity of the proposed method, we model two representative ground moving targets, i.e., person and wheeled vehicles, and carry out numerous comparative experiments. The result demonstrates that the developed framework can simultaneously recognize multiple targets and is also robust to variation of the signal-to-noise ratio (SNR), the initial position of targets, and the difference in scattering coefficient.},
keywords={},
doi={10.1587/transele.2022ECS6011},
ISSN={1745-1353},
month={August},}
Copier
TY - JOUR
TI - Multi-Target Recognition Utilizing Micro-Doppler Signatures with Limited Supervision
T2 - IEICE TRANSACTIONS on Electronics
SP - 454
EP - 457
AU - Jingyi ZHANG
AU - Kuiyu CHEN
AU - Yue MA
PY - 2023
DO - 10.1587/transele.2022ECS6011
JO - IEICE TRANSACTIONS on Electronics
SN - 1745-1353
VL - E106-C
IS - 8
JA - IEICE TRANSACTIONS on Electronics
Y1 - August 2023
AB - Previously, convolutional neural networks have made tremendous progress in target recognition based on micro-Doppler radar. However, these studies only considered the presence of one target at a time in the surveillance area. Simultaneous multi-targets recognition for surveillance radar remains a pretty challenging issue. To alleviate this issue, this letter develops a multi-instance multi-label (MIML) learning strategy, which can automatically locate the crucial input patterns that trigger the labels. Benefitting from its powerful target-label relation discovery ability, the proposed framework can be trained with limited supervision. We emphasize that only echoes from single targets are involved in training data, avoiding the preparation and annotation of multi-targets echo in the training stage. To verify the validity of the proposed method, we model two representative ground moving targets, i.e., person and wheeled vehicles, and carry out numerous comparative experiments. The result demonstrates that the developed framework can simultaneously recognize multiple targets and is also robust to variation of the signal-to-noise ratio (SNR), the initial position of targets, and the difference in scattering coefficient.
ER -