Multi-Source Domain Adaptation for Object Detection

被引:27
|
作者
Yao, Xingxu [1 ,3 ]
Zhao, Sicheng [2 ]
Xu, Pengfei [3 ]
Yang, Jufeng [1 ]
机构
[1] Nankai Univ, Tianjin, Peoples R China
[2] Columbia Univ, New York, NY USA
[3] Didi Chuxing, Beijing, Peoples R China
关键词
D O I
10.1109/ICCV48922.2021.00326
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
To reduce annotation labor associated with object detection, an increasing number of studies focus on transferring the learned knowledge from a labeled source domain to another unlabeled target domain. However, existing methods assume that the labeled data are sampled from a single source domain, which ignores a more generalized scenario, where labeled data are from multiple source domains. For the more challenging task, we propose a unified Faster R-CNN based framework, termed Divide-and-Merge Spindle Network (DMSN), which can simultaneously enhance domain invariance and preserve discriminative power. Specifically, the framework contains multiple source subnets and a pseudo target subnet. First, we propose a hierarchical feature alignment strategy to conduct strong and weak alignments for low- and high-level features, respectively, considering their different effects for object detection. Second, we develop a novel pseudo subnet learning algorithm to approximate optimal parameters of pseudo target subset by weighted combination of parameters in different source subnets. Finally, a consistency regularization for region proposal network is proposed to facilitate each subnet to learn more abstract invariances. Extensive experiments on different adaptation scenarios demonstrate the effectiveness of the proposed model.
引用
收藏
页码:3253 / 3262
页数:10
相关论文
共 50 条
  • [1] Multi-source unsupervised domain adaptation for object detection
    Zhang, Dan
    Ye, Mao
    Liu, Yiguang
    Xiong, Lin
    Zhou, Lihua
    [J]. INFORMATION FUSION, 2022, 78 : 138 - 148
  • [2] Weighted alignment-based multi-source domain adaptation for object detection
    Han, Joonhwan
    Woo, Seungbeom
    Hwang, Joong-won
    Hwang, Wonjun
    [J]. ELECTRONICS LETTERS, 2023, 59 (02)
  • [3] Multi-source domain adaptation of GPR data for IED detection
    Mehmet Oturak
    Seniha Esen Yuksel
    Sefa Kucuk
    [J]. Signal, Image and Video Processing, 2023, 17 : 1831 - 1839
  • [4] Multi-source domain adaptation of GPR data for IED detection
    Oturak, Mehmet
    Yuksel, Seniha Esen
    Kucuk, Sefa
    [J]. SIGNAL IMAGE AND VIDEO PROCESSING, 2023, 17 (05) : 1831 - 1839
  • [5] Multi-Source Distilling Domain Adaptation
    Zhao, Sicheng
    Wang, Guangzhi
    Zhang, Shanghang
    Gu, Yang
    Li, Yaxian
    Song, Zhichao
    Xu, Pengfei
    Hu, Runbo
    Chai, Hua
    Keutzer, Kurt
    [J]. THIRTY-FOURTH AAAI CONFERENCE ON ARTIFICIAL INTELLIGENCE, THE THIRTY-SECOND INNOVATIVE APPLICATIONS OF ARTIFICIAL INTELLIGENCE CONFERENCE AND THE TENTH AAAI SYMPOSIUM ON EDUCATIONAL ADVANCES IN ARTIFICIAL INTELLIGENCE, 2020, 34 : 12975 - 12983
  • [6] A survey of multi-source domain adaptation
    Sun, Shiliang
    Shi, Honglei
    Wu, Yuanbin
    [J]. INFORMATION FUSION, 2015, 24 : 84 - 92
  • [7] BAYESIAN MULTI-SOURCE DOMAIN ADAPTATION
    Sun, Shi-Liang
    Shi, Hong-Lei
    [J]. PROCEEDINGS OF 2013 INTERNATIONAL CONFERENCE ON MACHINE LEARNING AND CYBERNETICS (ICMLC), VOLS 1-4, 2013, : 24 - 28
  • [8] Multi-Source Survival Domain Adaptation
    Shaker, Ammar
    Lawrence, Carolin
    [J]. THIRTY-SEVENTH AAAI CONFERENCE ON ARTIFICIAL INTELLIGENCE, VOL 37 NO 8, 2023, : 9752 - 9762
  • [9] Building damage detection based on multi-source adversarial domain adaptation
    Wang, Xiang
    Li, Yundong
    Lin, Chen
    Liu, Yi
    Geng, Shuo
    [J]. JOURNAL OF APPLIED REMOTE SENSING, 2021, 15 (03)
  • [10] A weighted multi-source domain adaptation approach for surface defect detection
    Hu, Bing
    Wang, Jianhui
    [J]. IET IMAGE PROCESSING, 2022, 16 (08) : 2210 - 2218