Salient Object Detection Using Recurrent Guidance Network With Hierarchical Attention Features

Fully convolutional networks (FCNs) play an significant role in salient object detection tasks, due to the capability of extracting abundant multi-level and multi-scale features. However, most of FCN-based models utilize multi-level features in a single indiscriminative manner, which is difficult to...

Descripción completa

Guardado en:

Detalles Bibliográficos
Autores principales:	Shanmei Lu, Qiang Guo, Yongxia Zhang
Formato:	article
Lenguaje:	EN
Publicado:	IEEE 2020
Materias:	Salient object detection hierarchical features attention module recurrent network Electrical engineering. Electronics. Nuclear engineering TK1-9971
Acceso en línea:	https://doaj.org/article/0a89e388997c497fb40fe4beb2459347
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!

id	oai:doaj.org-article:0a89e388997c497fb40fe4beb2459347
record_format	dspace
spelling	oai:doaj.org-article:0a89e388997c497fb40fe4beb24593472021-11-19T00:05:55ZSalient Object Detection Using Recurrent Guidance Network With Hierarchical Attention Features2169-353610.1109/ACCESS.2020.3017512https://doaj.org/article/0a89e388997c497fb40fe4beb24593472020-01-01T00:00:00Zhttps://ieeexplore.ieee.org/document/9170598/https://doaj.org/toc/2169-3536Fully convolutional networks (FCNs) play an significant role in salient object detection tasks, due to the capability of extracting abundant multi-level and multi-scale features. However, most of FCN-based models utilize multi-level features in a single indiscriminative manner, which is difficult to accurately predict saliency maps. To address this problem, in this article, we propose a recurrent network which uses hierarchical attention features as a guidance for salient object detection. First of all, we divide multi-level features into low-level features and high-level features. Multi-scale features are extracted from high-level features using atrous convolutions with different receptive fields to obtain contextual information. Meanwhile, low-level features are refined as supplement to add detailed information in convolutional features. It is observed that the attention focus of hierarchical features is considerably different because of their distinct information representations. For this reason, a two-stage attention module is introduced for hierarchical features to guide the generation of saliency maps. Effective hierarchial attention features are obtained by aggregating the low-level and high-level features, but the attention of integrated features may be biased, leading to deviations in the detected salient regions. Therefore, we design a recurrent guidance network to correct the biased salient regions, which can effectively suppress the distractions in background and progressively refine salient objects boundaries. Experimental results show that our method exhibits superior performance in both quantitative and qualitative assessments on several widely used benchmark datasets.Shanmei LuQiang GuoYongxia ZhangIEEEarticleSalient object detectionhierarchical featuresattention modulerecurrent networkElectrical engineering. Electronics. Nuclear engineeringTK1-9971ENIEEE Access, Vol 8, Pp 151325-151334 (2020)
institution	DOAJ
collection	DOAJ
language	EN
topic	Salient object detection hierarchical features attention module recurrent network Electrical engineering. Electronics. Nuclear engineering TK1-9971
spellingShingle	Salient object detection hierarchical features attention module recurrent network Electrical engineering. Electronics. Nuclear engineering TK1-9971 Shanmei Lu Qiang Guo Yongxia Zhang Salient Object Detection Using Recurrent Guidance Network With Hierarchical Attention Features
description	Fully convolutional networks (FCNs) play an significant role in salient object detection tasks, due to the capability of extracting abundant multi-level and multi-scale features. However, most of FCN-based models utilize multi-level features in a single indiscriminative manner, which is difficult to accurately predict saliency maps. To address this problem, in this article, we propose a recurrent network which uses hierarchical attention features as a guidance for salient object detection. First of all, we divide multi-level features into low-level features and high-level features. Multi-scale features are extracted from high-level features using atrous convolutions with different receptive fields to obtain contextual information. Meanwhile, low-level features are refined as supplement to add detailed information in convolutional features. It is observed that the attention focus of hierarchical features is considerably different because of their distinct information representations. For this reason, a two-stage attention module is introduced for hierarchical features to guide the generation of saliency maps. Effective hierarchial attention features are obtained by aggregating the low-level and high-level features, but the attention of integrated features may be biased, leading to deviations in the detected salient regions. Therefore, we design a recurrent guidance network to correct the biased salient regions, which can effectively suppress the distractions in background and progressively refine salient objects boundaries. Experimental results show that our method exhibits superior performance in both quantitative and qualitative assessments on several widely used benchmark datasets.
format	article
author	Shanmei Lu Qiang Guo Yongxia Zhang
author_facet	Shanmei Lu Qiang Guo Yongxia Zhang
author_sort	Shanmei Lu
title	Salient Object Detection Using Recurrent Guidance Network With Hierarchical Attention Features
title_short	Salient Object Detection Using Recurrent Guidance Network With Hierarchical Attention Features
title_full	Salient Object Detection Using Recurrent Guidance Network With Hierarchical Attention Features
title_fullStr	Salient Object Detection Using Recurrent Guidance Network With Hierarchical Attention Features
title_full_unstemmed	Salient Object Detection Using Recurrent Guidance Network With Hierarchical Attention Features
title_sort	salient object detection using recurrent guidance network with hierarchical attention features
publisher	IEEE
publishDate	2020
url	https://doaj.org/article/0a89e388997c497fb40fe4beb2459347
work_keys_str_mv	AT shanmeilu salientobjectdetectionusingrecurrentguidancenetworkwithhierarchicalattentionfeatures AT qiangguo salientobjectdetectionusingrecurrentguidancenetworkwithhierarchicalattentionfeatures AT yongxiazhang salientobjectdetectionusingrecurrentguidancenetworkwithhierarchicalattentionfeatures
_version_	1718420649833136128

Salient Object Detection Using Recurrent Guidance Network With Hierarchical Attention Features

Ejemplares similares