Class structure‐aware adversarial loss for cross‐domain human action recognition

Abstract Cross‐domain action recognition is a challenging vision task due to the domain shift and the absence of labeled data in the target domain. With only labelled source domain and unlabelled target domain data during training, some existing methods rely on an adversarial framework to align the...

Descripción completa

Guardado en:

Detalles Bibliográficos
Autores principales:	Wanjun Chen, Long Liu, Guangfeng Lin, Yajun Chen, Jing Wang
Formato:	article
Lenguaje:	EN
Publicado:	Wiley 2021
Materias:	Photography TR1-1050 Computer software QA76.75-76.765
Acceso en línea:	https://doaj.org/article/f11746062f894edaa900ede431d6dd23
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!

id	oai:doaj.org-article:f11746062f894edaa900ede431d6dd23
record_format	dspace
spelling	oai:doaj.org-article:f11746062f894edaa900ede431d6dd232021-11-29T03:38:15ZClass structure‐aware adversarial loss for cross‐domain human action recognition1751-96671751-965910.1049/ipr2.12309https://doaj.org/article/f11746062f894edaa900ede431d6dd232021-12-01T00:00:00Zhttps://doi.org/10.1049/ipr2.12309https://doaj.org/toc/1751-9659https://doaj.org/toc/1751-9667Abstract Cross‐domain action recognition is a challenging vision task due to the domain shift and the absence of labeled data in the target domain. With only labelled source domain and unlabelled target domain data during training, some existing methods rely on an adversarial framework to align the features from different domains to a common latent space. However, the existing adversarial‐based approaches have a major limitation of only attempting to perform the alignment from a holistic view, ignoring the underlying coherence of class structure across domains. A class structure‐aware adversarial loss (CSCAL) is presented to address this issue. The CSCAL incorporates the category information into the adversarial learning branch to capture the fine‐grained alignment of each class, effectively avoiding the false mixup of samples from different categories in the embedding space. Experiments on HMDB51, UCF101 and Olympic Sports datasets show significant improvement compared to the baseline. Code and trained model can be found at https://github.com/bregmangh/CSCAL.Wanjun ChenLong LiuGuangfeng LinYajun ChenJing WangWileyarticlePhotographyTR1-1050Computer softwareQA76.75-76.765ENIET Image Processing, Vol 15, Iss 14, Pp 3425-3432 (2021)
institution	DOAJ
collection	DOAJ
language	EN
topic	Photography TR1-1050 Computer software QA76.75-76.765
spellingShingle	Photography TR1-1050 Computer software QA76.75-76.765 Wanjun Chen Long Liu Guangfeng Lin Yajun Chen Jing Wang Class structure‐aware adversarial loss for cross‐domain human action recognition
description	Abstract Cross‐domain action recognition is a challenging vision task due to the domain shift and the absence of labeled data in the target domain. With only labelled source domain and unlabelled target domain data during training, some existing methods rely on an adversarial framework to align the features from different domains to a common latent space. However, the existing adversarial‐based approaches have a major limitation of only attempting to perform the alignment from a holistic view, ignoring the underlying coherence of class structure across domains. A class structure‐aware adversarial loss (CSCAL) is presented to address this issue. The CSCAL incorporates the category information into the adversarial learning branch to capture the fine‐grained alignment of each class, effectively avoiding the false mixup of samples from different categories in the embedding space. Experiments on HMDB51, UCF101 and Olympic Sports datasets show significant improvement compared to the baseline. Code and trained model can be found at https://github.com/bregmangh/CSCAL.
format	article
author	Wanjun Chen Long Liu Guangfeng Lin Yajun Chen Jing Wang
author_facet	Wanjun Chen Long Liu Guangfeng Lin Yajun Chen Jing Wang
author_sort	Wanjun Chen
title	Class structure‐aware adversarial loss for cross‐domain human action recognition
title_short	Class structure‐aware adversarial loss for cross‐domain human action recognition
title_full	Class structure‐aware adversarial loss for cross‐domain human action recognition
title_fullStr	Class structure‐aware adversarial loss for cross‐domain human action recognition
title_full_unstemmed	Class structure‐aware adversarial loss for cross‐domain human action recognition
title_sort	class structure‐aware adversarial loss for cross‐domain human action recognition
publisher	Wiley
publishDate	2021
url	https://doaj.org/article/f11746062f894edaa900ede431d6dd23
work_keys_str_mv	AT wanjunchen classstructureawareadversariallossforcrossdomainhumanactionrecognition AT longliu classstructureawareadversariallossforcrossdomainhumanactionrecognition AT guangfenglin classstructureawareadversariallossforcrossdomainhumanactionrecognition AT yajunchen classstructureawareadversariallossforcrossdomainhumanactionrecognition AT jingwang classstructureawareadversariallossforcrossdomainhumanactionrecognition
_version_	1718407684364959744

Class structure‐aware adversarial loss for cross‐domain human action recognition

Ejemplares similares