Cross-Domain Person Re-Identification Based on Feature Fusion

Person re-identification (ReID) is one of the commonly used criminal investigation methods in reconnaissance. Although the current ReID has achieved robust results on single domains, the focus of researches has shifted to cross-domain in recent years, which is caused by domain bias between different...

Descripción completa

Guardado en:

Detalles Bibliográficos
Autores principales:	Xianjun Luo, Zhi Ouyang, Nisuo Du, Jingkuan Song, Qin Wei
Formato:	article
Lenguaje:	EN
Publicado:	IEEE 2021
Materias:	Person re-identification global feature local features semantic features feature erasure cross domain Electrical engineering. Electronics. Nuclear engineering TK1-9971
Acceso en línea:	https://doaj.org/article/e3d2d2756b5f429db396f9cc99483280
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!

id	oai:doaj.org-article:e3d2d2756b5f429db396f9cc99483280
record_format	dspace
spelling	oai:doaj.org-article:e3d2d2756b5f429db396f9cc994832802021-11-19T00:06:48ZCross-Domain Person Re-Identification Based on Feature Fusion2169-353610.1109/ACCESS.2021.3091647https://doaj.org/article/e3d2d2756b5f429db396f9cc994832802021-01-01T00:00:00Zhttps://ieeexplore.ieee.org/document/9462145/https://doaj.org/toc/2169-3536Person re-identification (ReID) is one of the commonly used criminal investigation methods in reconnaissance. Although the current ReID has achieved robust results on single domains, the focus of researches has shifted to cross-domain in recent years, which is caused by domain bias between different datasets. Generative Adversarial Networks (GAN) is used to realize the image style transfer of different datasets to alleviate the effect of cross-domain. However, the existing GAN-based models ignore complete expressions and occlusion of pedestrian characteristics, resulting in low accuracy in feature extraction. To address these issues, we introduce a cross domain model based on feature fusion (FFGAN) to fuse global, local and semantic features to extract more delicate pedestrian features. Before extracting pedestrian features, we preprocess feature maps with a feature erasure block to solve an occlusion issue. Finally, FFGAN enables a more complete visual description of pedestrian characteristics, thereby improving the accuracy of FFGAN in identifying pedestrians. Experimental results show that the effect of FFGAN is significantly improved compared with some advanced cross-domain ReID algorithms.Xianjun LuoZhi OuyangNisuo DuJingkuan SongQin WeiIEEEarticlePerson re-identificationglobal featurelocal featuressemantic featuresfeature erasurecross domainElectrical engineering. Electronics. Nuclear engineeringTK1-9971ENIEEE Access, Vol 9, Pp 98327-98336 (2021)
institution	DOAJ
collection	DOAJ
language	EN
topic	Person re-identification global feature local features semantic features feature erasure cross domain Electrical engineering. Electronics. Nuclear engineering TK1-9971
spellingShingle	Person re-identification global feature local features semantic features feature erasure cross domain Electrical engineering. Electronics. Nuclear engineering TK1-9971 Xianjun Luo Zhi Ouyang Nisuo Du Jingkuan Song Qin Wei Cross-Domain Person Re-Identification Based on Feature Fusion
description	Person re-identification (ReID) is one of the commonly used criminal investigation methods in reconnaissance. Although the current ReID has achieved robust results on single domains, the focus of researches has shifted to cross-domain in recent years, which is caused by domain bias between different datasets. Generative Adversarial Networks (GAN) is used to realize the image style transfer of different datasets to alleviate the effect of cross-domain. However, the existing GAN-based models ignore complete expressions and occlusion of pedestrian characteristics, resulting in low accuracy in feature extraction. To address these issues, we introduce a cross domain model based on feature fusion (FFGAN) to fuse global, local and semantic features to extract more delicate pedestrian features. Before extracting pedestrian features, we preprocess feature maps with a feature erasure block to solve an occlusion issue. Finally, FFGAN enables a more complete visual description of pedestrian characteristics, thereby improving the accuracy of FFGAN in identifying pedestrians. Experimental results show that the effect of FFGAN is significantly improved compared with some advanced cross-domain ReID algorithms.
format	article
author	Xianjun Luo Zhi Ouyang Nisuo Du Jingkuan Song Qin Wei
author_facet	Xianjun Luo Zhi Ouyang Nisuo Du Jingkuan Song Qin Wei
author_sort	Xianjun Luo
title	Cross-Domain Person Re-Identification Based on Feature Fusion
title_short	Cross-Domain Person Re-Identification Based on Feature Fusion
title_full	Cross-Domain Person Re-Identification Based on Feature Fusion
title_fullStr	Cross-Domain Person Re-Identification Based on Feature Fusion
title_full_unstemmed	Cross-Domain Person Re-Identification Based on Feature Fusion
title_sort	cross-domain person re-identification based on feature fusion
publisher	IEEE
publishDate	2021
url	https://doaj.org/article/e3d2d2756b5f429db396f9cc99483280
work_keys_str_mv	AT xianjunluo crossdomainpersonreidentificationbasedonfeaturefusion AT zhiouyang crossdomainpersonreidentificationbasedonfeaturefusion AT nisuodu crossdomainpersonreidentificationbasedonfeaturefusion AT jingkuansong crossdomainpersonreidentificationbasedonfeaturefusion AT qinwei crossdomainpersonreidentificationbasedonfeaturefusion
_version_	1718420641633271808

Cross-Domain Person Re-Identification Based on Feature Fusion

Ejemplares similares