Learning to Count in the Crowd from Limited Labeled Data

被引:47
作者
Sindagi, Vishwanath A. [1 ]
Yasarla, Rajeev [1 ]
Babu, Deepak Sam [2 ]
Babu, R. Venkatesh [2 ]
Patel, Vishal M. [1 ]
机构
[1] Johns Hopkins Univ, Baltimore, MD 21218 USA
[2] Indian Inst Sci, Bangalore 560012, India
来源
COMPUTER VISION - ECCV 2020, PT XI | 2020年 / 12356卷
关键词
Crowd counting; Semi-supervised learning; Pseudo-labeling; Domain adaptation; Synthetic to real transfer; PEOPLE;
D O I
10.1007/978-3-030-58621-8_13
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Recent crowd counting approaches have achieved excellent performance. However, they are essentially based on fully supervised paradigm and require large number of annotated samples. Obtaining annotations is an expensive and labour-intensive process. In this work, we focus on reducing the annotation efforts by learning to count in the crowd from limited number of labeled samples while leveraging a large pool of unlabeled data. Specifically, we propose a Gaussian Process-based iterative learning mechanism that involves estimation of pseudo-ground truth for the unlabeled data, which is then used as supervision for training the network. The proposed method is shown to be effective under the reduced data (semi-supervised) settings for several datasets like ShanghaiTech, UCF-QNRF, WorldExpo, UCSD, etc. Furthermore, we demonstrate that the proposed method can be leveraged to enable the network in learning to count from synthetic dataset while being able to generalize better to real-world datasets (synthetic-to-real transfer).
引用
收藏
页码:212 / 229
页数:18
相关论文
共 63 条
[1]  
Sindagi VA, 2020, Arxiv, DOI arXiv:2004.03597
[2]  
Sindagi VA, 2019, Arxiv, DOI arXiv:1907.10255
[3]   Counting in the Wild [J].
Arteta, Carlos ;
Lempitsky, Victor ;
Zisserman, Andrew .
COMPUTER VISION - ECCV 2016, PT VII, 2016, 9911 :483-498
[4]  
Sam DB, 2020, Arxiv, DOI arXiv:1906.07538
[5]   Scale Aggregation Network for Accurate and Efficient Crowd Counting [J].
Cao, Xinkun ;
Wang, Zhipeng ;
Zhao, Yanyun ;
Su, Fei .
COMPUTER VISION - ECCV 2018, PT V, 2018, 11209 :757-773
[6]   Privacy preserving crowd monitoring: Counting people without people models or tracking [J].
Chan, Antoni B. ;
Liang, Zhang-Sheng John ;
Vasconcelos, Nuno .
2008 IEEE CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION, VOLS 1-12, 2008, :1766-1772
[7]   Bayesian Poisson Regression for Crowd Counting [J].
Chan, Antoni B. ;
Vasconcelos, Nuno .
2009 IEEE 12TH INTERNATIONAL CONFERENCE ON COMPUTER VISION (ICCV), 2009, :545-551
[8]   Feature Mining for Localised Crowd Counting [J].
Chen, Ke ;
Loy, Chen Change ;
Gong, Shaogang ;
Xiang, Tao .
PROCEEDINGS OF THE BRITISH MACHINE VISION CONFERENCE 2012, 2012,
[9]  
Gao GS, 2020, Arxiv, DOI [arXiv:2003.12783, DOI 10.6705/J.JACME.202303_13(1).0002, 10.48550/arXiv.2003.12783]
[10]   Composition Loss for Counting, Density Map Estimation and Localization in Dense Crowds [J].
Idrees, Haroon ;
Tayyab, Muhmmad ;
Athrey, Kishan ;
Zhang, Dong ;
Al-Maadeed, Somaya ;
Rajpoot, Nasir ;
Shah, Mubarak .
COMPUTER VISION - ECCV 2018, PT II, 2018, 11206 :544-559