Source-free domain adaptive object detection based on pseudo-supervised mean teacher

doi:10.1007/s11227-022-04915-4

	Source-free domain adaptive object detection based on pseudo-supervised mean teacher
	Wei, Xing 1,3,4; Bai, Ting 1; Zhai, Yan 1; Chen, Lei 5; Luo, Hui 3; Zhao, Chong 2,3; Lu, Yang 1,4
刊名	JOURNAL OF SUPERCOMPUTING
	2022-11-02
关键词	Source-free object detection Transfer learning Domain adaptation
ISSN号	0920-8542
DOI	10.1007/s11227-022-04915-4
通讯作者	Bai, Ting(baiting@mail.hfut.edu.cn)
英文摘要	Domain adaptive object detection refers to training a cross-domain object detector through a large number of labeled source domain datasets and unlabeled target domain datasets and learning the domain invariant features between two domains to reduce or eliminate the domain discrepancy. However, factors such as data privacy protection, limited storage space, and high labor costs often make many source domain-labeled samples unavailable in real-time situations. In this work, we propose a pseudo-supervised mean teacher model for source-free domain adaptive object detection that alternates between generating pseudo-labels and fine-tuning the model and utilizes a pixel-level distillation loss method and the weight regularization module for model adaptation. We use the mean teacher model to assist training to achieve object detection task in the source-free domain. Experiments are carried out on multiple datasets such as Cityscapes, Foggy Cityscapes, and SIM10K. Extensive experiments on multiple domain adaptation scenarios show that our method achieves better performance than the baseline (Faster R-CNN) and multiple state-of-the-art domain adaptation methods which require access to source domain data, demonstrating the effectiveness and robustness of the proposed method.
资助项目	Joint Fund of Natural Science Foundation of Anhui Province[2008085UD08] ; Anhui Provincial Key R D Program[202004a05020004] ; Open fund of Intelligent Interconnected Systems Laboratory of Anhui Province[PA2021AKSK0107] ; Intelligent Networking and New Energy Vehicle Special Project of Intelligent Manufacturing Institute of HFUT[IMIWL2019003] ; Intelligent Networking and New Energy Vehicle Special Project of Intelligent Manufacturing Institute of HFUT[IMIDC2019002]
WOS研究方向	Computer Science ; Engineering
语种	英语
出版者	SPRINGER
WOS记录号	WOS:000878004600001
资助机构	Joint Fund of Natural Science Foundation of Anhui Province ; Anhui Provincial Key R D Program ; Open fund of Intelligent Interconnected Systems Laboratory of Anhui Province ; Intelligent Networking and New Energy Vehicle Special Project of Intelligent Manufacturing Institute of HFUT
内容类型	期刊论文
源URL	[http://ir.hfcas.ac.cn:8080/handle/334002/129993]
专题	中国科学院合肥物质科学研究院
通讯作者	Bai, Ting
作者单位	1.Hefei Univ Technol, Sch Comp & Informat, Emerald Rd 420, Hefei 230601, Anhui, Peoples R China 2.Hefei Univ Technol, Engn Qual Educ Ctr, Undergrad Sch, Hefei 230601, Anhui, Peoples R China 3.Hefei Univ Technol, Intelligent Mfg Inst, Emerald Rd 420, Hefei 230051, Anhui, Peoples R China 4.Minist Educ, Engn Res Ctr Safety Crit Ind Measurement & Contro, Hefei 230009, Anhui, Peoples R China 5.Chinese Acad Sci, Inst Intelligent Machines, Hefei 230031, Anhui, Peoples R China
推荐引用方式 GB/T 7714	Wei, Xing,Bai, Ting,Zhai, Yan,et al. Source-free domain adaptive object detection based on pseudo-supervised mean teacher[J]. JOURNAL OF SUPERCOMPUTING,2022.
APA	Wei, Xing.,Bai, Ting.,Zhai, Yan.,Chen, Lei.,Luo, Hui.,...&Lu, Yang.(2022).Source-free domain adaptive object detection based on pseudo-supervised mean teacher.JOURNAL OF SUPERCOMPUTING.
MLA	Wei, Xing,et al."Source-free domain adaptive object detection based on pseudo-supervised mean teacher".JOURNAL OF SUPERCOMPUTING (2022).