一种新的基于卷积神经网络的人群计数方法

黄杰浩; 遆晓光; 吴俊德; 陈瑷玥

doi:10.1631/FITEE.1900282

Your Location：

Home >

Browse articles >

一种新的基于卷积神经网络的人群计数方法

常规文章 | Updated：2022-05-19

- 一种新的基于卷积神经网络的人群计数方法
- A novel convolutional neural network method for crowd counting
- 信息与电子工程前沿（英文） 2020年21卷第8期页码：1150-1160
- Affiliations：
  
  Control and Simulation Center, Harbin Institute of Technology, Harbin 150080, China
- Author bio：
  
  [ "Jie-hao HUANG, E-mail: 18s004055@hit.edu.cn" ]
  Xiao-guang DI, E-mail: dixiaoguang@hit.edu.cn
- Funds：
  
  Project supported by the National Natural Science Foundation of China (No. 61775048) and the Fundamental Research Funds for the Central Universities, China (No. ZDXMPY20180103)
- DOI：10.1631/FITEE.1900282
  中图分类号： TP391.4
- 收稿：2019-06-05，
  
  修回：2020-;6-11，
  
  纸质出版：2020-08
- Accepted：
Scan QR Code
黄杰浩, 遆晓光, 吴俊德, 等. 一种新的基于卷积神经网络的人群计数方法[J]. 信息与电子工程前沿（英文）, 2020,21(8):1150-1160.

Jie-hao HUANG, Xiao-guang DI, Jun-de WU, et al. A novel convolutional neural network method for crowd counting[J]. Frontiers of Information Technology & Electronic Engineering, 2020, 21(8): 1150-1160.
黄杰浩, 遆晓光, 吴俊德, 等. 一种新的基于卷积神经网络的人群计数方法[J]. 信息与电子工程前沿（英文）, 2020,21(8):1150-1160. DOI： 10.1631/FITEE.1900282.

Jie-hao HUANG, Xiao-guang DI, Jun-de WU, et al. A novel convolutional neural network method for crowd counting[J]. Frontiers of Information Technology & Electronic Engineering, 2020, 21(8): 1150-1160. DOI： 10.1631/FITEE.1900282.

摘要

人群密度估计是一项具有挑战性的任务，因为人群中人头大小存在大范围变化。现有方法均采用多列式结构卷积神经网络去适应这种变化，但会导致密度图上不同密度区域产生平均效应并引入额外噪声。为解决该问题，提出一种新的基于分割先验图的神经网络方法，在分割图基础上生成一个高质量且没有噪声的密度图。该网络主要包括两个部分，即头部的人群前景分割神经网络和尾部的人群回归神经网络。在数据集只提供单点人头标记的情况下，采用均匀函数生成人群头部的掩膜真值图。基于该真值图，前景分割网络输出人群分割图，可有效减少密度图中无人区域噪声。将人群分割图输入人群回归网络，后者能生成高质量人群密度图并提供准确的人数估计。在4个公开数据集（即ShanghaiTech、UCF-CC-50、WorldExpo’10和UCSD）上验证了所提方法有效性；其中，在ShanghaiTech partB和UCF-CC-50两个数据集上该方法取得了当前最好结果。

Abstract

Crowd density estimation

in general

is a challenging task due to the large variation of head sizes in the crowds. Existing methods always use a multi-column convolutional neural network (MCNN) to adapt to this variation

which results in an average effect in areas with different densities and brings a lot of noise to the density map. To address this problem

we propose a new method called the segmentation-aware prior network (SAPNet)

which generates a high-quality density map without noise based on a coarse head-segmentation map. SAPNet is composed of two networks

i.e.

a foreground-segmentation convolutional neural network (FS-CNN) as the front end and a crowd-regression convolutional neural network (CR-CNN) as the back end. With only the single dot annotation

we generate the ground truth of segmentation masks in heads. Then

based on the ground truth

FS-CNN outputs a coarse head-segmentation map

which helps eliminate the noise in regions without people in the density map. By inputting the head-segmentation map generated by the front end

CR-CNN performs accurate crowd counting estimation and generates a high-quality density map. We demonstrate SAPNet on four datasets (i.e.

ShanghaiTech

UCF-CC-50

WorldExpo'10

and UCSD)

and show the state-of-the-art performances on ShanghaiTech part

and UCF-CC-50 datasets.

关键词

Keywords

references

J Canny . . A computational approach to edge detection . . IEEE Trans Patt Anal Mach Intell , , 1986 . . 8 ( ( 6 ): ): 679 - - 698 . . DOI: 10.1109/TPAMI.1986.4767851 http://doi.org/10.1109/TPAMI.1986.4767851 . .

AB Chan , , , N Vasconcelos . . Bayesian Poisson regression for crowd counting . . Proc IEEE 12 th Int Conf on Computer Vision , , 2009 . . p.545 - - 551 . . DOI: 10.1109/ICCV.2009.5459191 http://doi.org/10.1109/ICCV.2009.5459191 . .

AB Chan , , , ZSJ Liang , , , N Vasconcelos . . Privacy preserving crowd monitoring: counting people without people models or tracking . . Proc IEEE Conf on Computer Vision and Pattern Recognition , , 2008 . . p.1 - - 7 . . DOI: 10.1109/CVPR.2008.4587569 http://doi.org/10.1109/CVPR.2008.4587569 . .

JF Dai , , , Y Li , , , KM He , , , 等 . . R-FCN: object detection via region-based fully convolutional networks . . Proc 30 th Int Conf on Neural Information Processing Systems , , 2016 . . p.379 - - 387 . . . .

P Dollar , , , C Wojek , , , B Schiele , , , 等 . . Pedestrian detection: an evaluation of the state of the art . . IEEE Trans Patt Anal Mach Intell , , 2012 . . 34 ( ( 4 ): ): 743 - - 761 . . DOI: 10.1109/TPAMI.2011.155 http://doi.org/10.1109/TPAMI.2011.155 . .

H Idrees , , , I Saleemi , , , C Seibert , , , 等 . . Multi-source multi-scale counting in extremely dense crowd images . . Proc IEEE Conf on Computer Vision and Pattern Recognition , , 2013 . . p.2547 - - 2554 . . DOI: 10.1109/CVPR.2013.329 http://doi.org/10.1109/CVPR.2013.329 . .

K Kang , , , XG Wang . . Fully convolutional neural networks for crowd segmentation . . 2014 . . https://arxiv.org/abs/1411.4464 https://arxiv.org/abs/1411.4464 , , . .

V Lempitsky , , , A Zisserman . . Learning to count objects in images . . Proc 23 rd Int Conf on Neural Information Processing Systems , , 2010 . . p.1324 - - 1332 . . . .

HH Li , , , XJ He , , , HF Wu , , , 等 . . Structured inhomogeneous density map learning for crowd counting . . 2018 . . https://arxiv.org/abs/1801.06642 https://arxiv.org/abs/1801.06642 , , . .

JJ Li , , , H Yang , , , S Wu . . Crowd semantic segmentation based on spatial-temporal dynamics . . Proc 13 th IEEE Int Conf on Advanced Video and Signal Based Surveillance , , 2016 . . p.102 - - 108 . . DOI: 10.1109/AVSS.2016.7738032 http://doi.org/10.1109/AVSS.2016.7738032 . .

T Li , , , H Chang , , , M Wang , , , 等 . . Crowded scene analysis: a survey . . IEEE Trans Circ Syst Video Technol , , 2015 . . 25 ( ( 3 ): ): 367 - - 386 . . DOI: 10.1109/TCSVT.2014.2358029 http://doi.org/10.1109/TCSVT.2014.2358029 . .

YH Li , , , XF Zhang , , , DM Chen . . CSRNet: dilated convolutional neural networks for understanding the highly congested scenes . . Proc IEEE/CVF Conf on Computer Vision and Pattern Recognition , , 2018 . . p.1091 - - 1100 . . DOI: 10.1109/CVPR.2018.00120 http://doi.org/10.1109/CVPR.2018.00120 . .

J Liu , , , CQ Gao , , , DY Meng , , , 等 . . DecideNet: counting varying density crowds through attention guided detection and density estimation . . Proc IEEE/CVF Conf on Computer Vision and Pattern Recognition , , 2018 . . p.5197 - - 5206 . . DOI: 10.1109/CVPR.2018.00545 http://doi.org/10.1109/CVPR.2018.00545 . .

J Long , , , E Shelhamer , , , T Darrell . . Fully convolutional networks for semantic segmentation . . Proc IEEE Conf on Computer Vision and Pattern Recognition , , 2015 . . p.3431 - - 3440 . . DOI: 10.1109/CVPR.2015.7298965 http://doi.org/10.1109/CVPR.2015.7298965 . .

DB Sam , , , S Surya , , , RV Babu . . Switching convolutional neural network for crowd counting . . Proc IEEE Conf on Computer Vision and Pattern Recognition , , 2017 . . p.4031 - - 4039 . . DOI: 10.1109/CVPR.2017.429 http://doi.org/10.1109/CVPR.2017.429 . .

DB Sam , , , NN Sajjan , , , RV Babu . . Divide and grow: capturing huge diversity in crowd images with incrementally growing CNN . . Proc IEEE/CVF Conf on Computer Vision and Pattern Recognition , , 2018 . . p.3618 - - 3626 . . DOI: 10.1109/CVPR.2018.00381 http://doi.org/10.1109/CVPR.2018.00381 . .

Z Shen , , , Y Xu , , , B Ni , , , 等 . . Crowd counting via adversarial cross-scale consistency pursuit . . IEEE/CVF Conf on Computer Vision and Pattern Recognition , , 2018 . . p.5245 - - 5254 . . DOI: 10.1109/CVPR.2018.00550 http://doi.org/10.1109/CVPR.2018.00550 . .

VA Sindagi , , , VM Patel . . Generating high-quality crowd density maps using contextual pyramid CNNs . . Proc IEEE Int Conf on Computer Vision , , 2017 . . p.1879 - - 1888 . . DOI: 10.1109/ICCV.2017.206 http://doi.org/10.1109/ICCV.2017.206 . .

VA Sindagi , , , VM Patel . . A survey of recent advances in CNN-based single image crowd counting and density estimation . . Patt Recogn Lett , , 2018 . . 107 3 - - 16 . . DOI: 10.1016/j.patrec.2017.07.007 http://doi.org/10.1016/j.patrec.2017.07.007 . .

BB Zhan , , , DN Monekosso , , , P Remagnino , , , 等 . . Crowd analysis: a survey . . Mach Vis Appl , , 2008 . . 19 ( ( 5-6 ): ): 345 - - 357 . . DOI: 10.1007/s00138-008-0132-4 http://doi.org/10.1007/s00138-008-0132-4 . .

C Zhang , , , HS Li , , , XG Wang , , , 等 . . Cross-scene crowd counting via deep convolutional neural networks . . Proc IEEE Conf on Computer Vision and Pattern Recognition , , 2015 . . p.833 - - 841 . . DOI: 10.1109/CVPR.2015.7298684 http://doi.org/10.1109/CVPR.2015.7298684 . .

C Zhang , , , K Zhang , , , HS Li , , , 等 . . Data-driven crowd understanding: a baseline for a large-scale crowd dataset . . IEEE Trans Multim , , 2016 . . 18 ( ( 6 ): ): 1048 - - 1061 . . DOI: 10.1109/TMM.2016.2542585 http://doi.org/10.1109/TMM.2016.2542585 . .

YY Zhang , , , DS Zhou , , , SQ Chen , , , 等 . . Single-image crowd counting via multi-column convolutional neural network . . Proc IEEE Conf on Computer Vision and Pattern Recognition , , 2016 . . p.589 - - 597 . . DOI: 10.1109/CVPR.2016.70 http://doi.org/10.1109/CVPR.2016.70 . .

浏览量

Downloads

CSCD

文章被引用时，请邮件提醒。

Submit

工具集

关联资源

Aggregated context network for crowd counting

Forget less, count better: a domain-incremental self-distillation learning benchmark for lifelong crowd counting