Crowd counting with limited labeling through submodular frame selection

Qi Zhou, Junping Zhang, Lingfu Che, Hongming Shan, James Wang

Research output: Contribution to journalArticle

2 Citations (Scopus)

Abstract

Automated crowd counting is valuable for intelligent transportation systems, as it can help to improve the emergency planning and prevent congestion in transit hubs such as train stations and airports. Semi-supervised crowd counting aims to estimate the number of pedestrians in an ongoing scene using a combination of a small number of labeled frames and a large number of unlabeled ones. However, existing methods do not incorporate ways to effectively select informative frames as labeled training samples, resulting in low accuracy on unseen crowd scenes. We propose a submodular method to select the most informative frames from the image sequences of crowds. Specifically, the method selects the most representative images to guarantee the information coverage, by maximizing the similarities between the group of selected images and the image sequence. In addition, these frames are chosen to avoid redundancies and preserve diversity. Finally, our semi-supervised method incorporates graph Laplacian regularization and spatiotemporal constraints. Extensive experiments on three benchmark data sets demonstrate that our proposed approach achieves higher accuracy compared with the state-of-the-art regression methods and competitive performance with deep convolutional models, especially when the number of labeled data is exceptionally small.

Original languageEnglish (US)
Article number8360780
Pages (from-to)1728-1738
Number of pages11
JournalIEEE Transactions on Intelligent Transportation Systems
Volume20
Issue number5
DOIs
StatePublished - May 1 2019

Fingerprint

Airports
Labeling
Redundancy
Planning
Experiments

All Science Journal Classification (ASJC) codes

  • Automotive Engineering
  • Mechanical Engineering
  • Computer Science Applications

Cite this

Zhou, Qi ; Zhang, Junping ; Che, Lingfu ; Shan, Hongming ; Wang, James. / Crowd counting with limited labeling through submodular frame selection. In: IEEE Transactions on Intelligent Transportation Systems. 2019 ; Vol. 20, No. 5. pp. 1728-1738.
@article{dbc017de80e1484f88b09602816fcb3d,
title = "Crowd counting with limited labeling through submodular frame selection",
abstract = "Automated crowd counting is valuable for intelligent transportation systems, as it can help to improve the emergency planning and prevent congestion in transit hubs such as train stations and airports. Semi-supervised crowd counting aims to estimate the number of pedestrians in an ongoing scene using a combination of a small number of labeled frames and a large number of unlabeled ones. However, existing methods do not incorporate ways to effectively select informative frames as labeled training samples, resulting in low accuracy on unseen crowd scenes. We propose a submodular method to select the most informative frames from the image sequences of crowds. Specifically, the method selects the most representative images to guarantee the information coverage, by maximizing the similarities between the group of selected images and the image sequence. In addition, these frames are chosen to avoid redundancies and preserve diversity. Finally, our semi-supervised method incorporates graph Laplacian regularization and spatiotemporal constraints. Extensive experiments on three benchmark data sets demonstrate that our proposed approach achieves higher accuracy compared with the state-of-the-art regression methods and competitive performance with deep convolutional models, especially when the number of labeled data is exceptionally small.",
author = "Qi Zhou and Junping Zhang and Lingfu Che and Hongming Shan and James Wang",
year = "2019",
month = "5",
day = "1",
doi = "10.1109/TITS.2018.2829987",
language = "English (US)",
volume = "20",
pages = "1728--1738",
journal = "IEEE Transactions on Intelligent Transportation Systems",
issn = "1524-9050",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
number = "5",

}

Crowd counting with limited labeling through submodular frame selection. / Zhou, Qi; Zhang, Junping; Che, Lingfu; Shan, Hongming; Wang, James.

In: IEEE Transactions on Intelligent Transportation Systems, Vol. 20, No. 5, 8360780, 01.05.2019, p. 1728-1738.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Crowd counting with limited labeling through submodular frame selection

AU - Zhou, Qi

AU - Zhang, Junping

AU - Che, Lingfu

AU - Shan, Hongming

AU - Wang, James

PY - 2019/5/1

Y1 - 2019/5/1

N2 - Automated crowd counting is valuable for intelligent transportation systems, as it can help to improve the emergency planning and prevent congestion in transit hubs such as train stations and airports. Semi-supervised crowd counting aims to estimate the number of pedestrians in an ongoing scene using a combination of a small number of labeled frames and a large number of unlabeled ones. However, existing methods do not incorporate ways to effectively select informative frames as labeled training samples, resulting in low accuracy on unseen crowd scenes. We propose a submodular method to select the most informative frames from the image sequences of crowds. Specifically, the method selects the most representative images to guarantee the information coverage, by maximizing the similarities between the group of selected images and the image sequence. In addition, these frames are chosen to avoid redundancies and preserve diversity. Finally, our semi-supervised method incorporates graph Laplacian regularization and spatiotemporal constraints. Extensive experiments on three benchmark data sets demonstrate that our proposed approach achieves higher accuracy compared with the state-of-the-art regression methods and competitive performance with deep convolutional models, especially when the number of labeled data is exceptionally small.

AB - Automated crowd counting is valuable for intelligent transportation systems, as it can help to improve the emergency planning and prevent congestion in transit hubs such as train stations and airports. Semi-supervised crowd counting aims to estimate the number of pedestrians in an ongoing scene using a combination of a small number of labeled frames and a large number of unlabeled ones. However, existing methods do not incorporate ways to effectively select informative frames as labeled training samples, resulting in low accuracy on unseen crowd scenes. We propose a submodular method to select the most informative frames from the image sequences of crowds. Specifically, the method selects the most representative images to guarantee the information coverage, by maximizing the similarities between the group of selected images and the image sequence. In addition, these frames are chosen to avoid redundancies and preserve diversity. Finally, our semi-supervised method incorporates graph Laplacian regularization and spatiotemporal constraints. Extensive experiments on three benchmark data sets demonstrate that our proposed approach achieves higher accuracy compared with the state-of-the-art regression methods and competitive performance with deep convolutional models, especially when the number of labeled data is exceptionally small.

UR - http://www.scopus.com/inward/record.url?scp=85047209585&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85047209585&partnerID=8YFLogxK

U2 - 10.1109/TITS.2018.2829987

DO - 10.1109/TITS.2018.2829987

M3 - Article

AN - SCOPUS:85047209585

VL - 20

SP - 1728

EP - 1738

JO - IEEE Transactions on Intelligent Transportation Systems

JF - IEEE Transactions on Intelligent Transportation Systems

SN - 1524-9050

IS - 5

M1 - 8360780

ER -