Dilated Convolution-based Feature Refinement Network for Crowd Localization

doi:10.1145/3571134

CORC > 自动化研究所 > 中国科学院自动化研究所 > 多模态人工智能系统全国重点实验室

	Dilated Convolution-based Feature Refinement Network for Crowd Localization
	Gao, Xingyu 7; Xie, Jinyang 6; Chen, Zhenyu 1,5; Liu, An-An 4; Sun, Zhenan 2,3; Lyu, Lei 6
刊名	ACM TRANSACTIONS ON MULTIMEDIA COMPUTING COMMUNICATIONS AND APPLICATIONS
	2023-11-01
卷号	19 期号:6 页码:16
关键词	Dilated convolution Feature Refinement crowd localization contextual information
ISSN号	1551-6857
DOI	10.1145/3571134
通讯作者	Xie, Jinyang(xiejinyangsdnu@163.com) ; Lyu, Lei(lvlei@sdnu.edu.cn)
英文摘要	As an emerging computer vision task, crowd localization has received increasing attention due to its ability to produce more accurate spatially predictions. However, continuous scale variations in complex crowd scenes lead to tiny individuals at the edges, so that existing methods cannot achieve precise crowd localization. Aiming at alleviating the above problems, we propose a novel Dilated Convolution-based Feature Refinement Network (DFRNet) to enhance the representation learning capability. Specifically, the DFRNet is built with three branches that can capture the information of each individual in crowd scenes more precisely. More specifically, we introduce a Feature Perception Module to model long-range contextual information at different scales by adopting multiple dilated convolutions, thus providing sufficient feature information to perceive tiny individuals at the edge of images. Afterwards, a Feature Refinement Module is deployed at multiple stages of the three branches to facilitate the mutual refinement of feature information at different scales, thus further improving the expression capability of multi-scale contextual information. By incorporating the above modules, DFRNet can locate individuals in complex scenes more precisely. Extensive experiments on multiple datasets demonstrate that the proposed method has more advanced performance compared to existing methods and can be more accurately adapted to complex crowd scenes.
资助项目	National Natural Science Foundation of China[61976127] ; Science and Technology Innovation 2030-Major Project (Brain Science and Brain-Like Intelligence Technology)[2022ZD0208700]
WOS关键词	MEAN SQUARED ERROR
WOS研究方向	Computer Science
语种	英语
出版者	ASSOC COMPUTING MACHINERY
WOS记录号	WOS:001035785200039
资助机构	National Natural Science Foundation of China ; Science and Technology Innovation 2030-Major Project (Brain Science and Brain-Like Intelligence Technology)
内容类型	期刊论文
源URL	[http://ir.ia.ac.cn/handle/173211/54002]
专题	多模态人工智能系统全国重点实验室
通讯作者	Xie, Jinyang; Lyu, Lei
作者单位	1.State Grid Corp China, Big Data Ctr, Beijing, Peoples R China 2.Univ Chinese Acad Sci, Beijing, Peoples R China 3.Chinese Acad Sci, Inst Automat, Beijing, Peoples R China 4.Tianjin Univ, Sch Elect & Informat Engn, Tianjin, Peoples R China 5.China Elect Power Res Inst, Beijing, Peoples R China 6.Shandong Normal Univ, Sch Informat Sci & Engn, Jinan, Peoples R China 7.Chinese Acad Sci, Inst Microelect, Beijing, Peoples R China
推荐引用方式 GB/T 7714	Gao, Xingyu,Xie, Jinyang,Chen, Zhenyu,et al. Dilated Convolution-based Feature Refinement Network for Crowd Localization[J]. ACM TRANSACTIONS ON MULTIMEDIA COMPUTING COMMUNICATIONS AND APPLICATIONS,2023,19(6):16.
APA	Gao, Xingyu,Xie, Jinyang,Chen, Zhenyu,Liu, An-An,Sun, Zhenan,&Lyu, Lei.(2023).Dilated Convolution-based Feature Refinement Network for Crowd Localization.ACM TRANSACTIONS ON MULTIMEDIA COMPUTING COMMUNICATIONS AND APPLICATIONS,19(6),16.
MLA	Gao, Xingyu,et al."Dilated Convolution-based Feature Refinement Network for Crowd Localization".ACM TRANSACTIONS ON MULTIMEDIA COMPUTING COMMUNICATIONS AND APPLICATIONS 19.6(2023):16.