融合注意力机制与上下文密度图的人群计数网络

doi:10.19678/j.issn.1000-3428.0063039

摘要/Abstract

摘要： 为分析商业区人群流动情况，或避免人群踩踏等公共事件的发生，通常采用人群计数方法统计监控图像中的人数信息，从而达到提前预警的效果。受目标遮挡、背景干扰、多尺度变化等因素的影响，现有的人群计数方法在统计人数信息的过程中存在误算或漏算的问题，导致准确率降低。提出一种基于注意力机制与上下文密度图融合的人群计数网络CADMFNet。以VGG16的部分卷积层作为前端网络，通过引入上采样融合模块对输入的特征图进行上下文特征融合，将不同膨胀率的膨胀卷积作为后端网络，生成高质量的中间密度图。在此基础上，采用上下文注意力模块融合不同层级的中间密度图，获得精细的人群密度图。实验结果表明，该网络在Mall数据集上的平均绝对误差和均方根误差分别为1.31和1.59，相比CSRNet、MCNN等网络，能够有效提高计数的准确度，并且具有较优的鲁棒性。

关键词: 人群计数, 特征融合, 膨胀卷积, 注意力机制, 卷积神经网络

Abstract: To analyze the flow of people in a business district and avoid crowd stampedes, the crowd counting method is usually used to count the number of people in a monitoring image to achieve the effect of early warning.Affected by target occlusion, background interference, multi-scale change, and other factors, existing crowd counting methods have miscalculation and omission in the process of counting the number of people, resulting in low accuracy.A crowd counting network, CADMFNet, based on attention mechanism and contextual density map fusion is proposed.Taking part of the convolution layer of VGG16 as the front-end network, the contextual features of the input feature map are fused by introducing the up-sampling fusion module, and the dilated convolution with different dilation rates are used as the back-end network to generate a high-quality intermediate density map.On this basis, the contextual attention module is used to fuse the intermediate density maps of different levels to obtain a fine population density map.The experimental results show that the average absolute error and root mean square error of the network on the Mall data set are 1.31 and 1.59, respectively.Compared with CSRNet, MCNN, and other networks, it effectively improves the counting accuracy and exhibited better robustness.

Key words: crowd counting, feature fusion, dilated convolution, attention mechanism, Convolutional Neural Network (CNN)

中图分类号:

TP391

吴奇元, 王晓东, 章联军, 高海玲, 赵伸豪. 融合注意力机制与上下文密度图的人群计数网络[J]. 计算机工程, 2022, 48(5): 235-241,250.

WU Qiyuan, WANG Xiaodong, ZHANG Lianjun, GAO Hailing, ZHAO Shenhao. Crowd Counting Network with Attention Mechanism and Contextual Density Map[J]. Computer Engineering, 2022, 48(5): 235-241,250.

http://www.ecice06.com/CN/Y2022/V48/I5/235

图/表 10

20220806174515

20220806174518

20220806174522

20220806174526

20220806174534

20220806174538

20220806174542

20220806174545

20220806174557

20220806174600

参考文献

[1] ZOU L H, LIU Y C.A new algorithm of counting human based on segmentation of human faces in color image[C]//Proceedings of International Conference on Computational Intelligence and Security.Washington D.C., USA:IEEE Press, 2009:505-509.
[2] HOU Y L, PANG G K H.People counting and human detection in a challenging situation[J].IEEE Transactions on Systems, Man, and Cybernetics-Part A:Systems and Humans, 2011, 41(1):24-33.
[3] IDREES H, SOOMRO K, SHAH M.Detecting humans in dense crowds using locally-consistent scale prior and global occlusion reasoning[J].IEEE Transactions on Pattern Analysis and Machine Intelligence, 2015, 37(10):1986-1998.
[4] RYAN D, DENMAN S, FOOKES C, et al.Crowd counting using multiple local features[C]//Proceedings of Digital Image Computing:Techniques and Applications.Washington D.C., USA:IEEE Press, 2009:81-88.
[5] CHEN K, LOY C C, GONG S G, et al.Feature mining for localised crowd counting[EB/OL].[2021-09-22].http://www.eecs.qmul.ac.uk/~sgg/papers/ChenEtAl_BMVC2012.pdf.
[6] CHAN A B, VASCONCELOS N.Bayesian poisson regression for crowd counting[C]//Proceedings of the 12th International Conference on Computer Vision.Washington D.C., USA:IEEE Press, 2009:545-551.
[7] LEMPITSKY V, ZISSERMAN A.Learning to count objects in images[C]//Proceedings of the 23rd International Conference on Neural Information Processing Systems.Cambridge, USA:MIT Press, 2010:1324-1332.
[8] ZHANG Y Y, ZHOU D S, CHEN S Q, et al.Single-image crowd counting via multi-column convolutional neural network[C]//Proceedings of IEEE Conference on Computer Vision and Pattern Recognition.Washington D.C., USA:IEEE Press, 2016:589-597.
[9] SAM D B, SURYA S, BABU R V.Switching convolutional neural network for crowd counting[C]//Proceedings of IEEE Conference on Computer Vision and Pattern Recognition.Washington D.C., USA:IEEE Press, 2017:4031-4039.
[10] 陆金刚, 张莉.基于多尺度多列卷积神经网络的密集人群计数模型[J].计算机应用, 2019, 39(12):3445-3449. LU J G, ZHANG L.Crowd counting model based on multi-scale multi-column convolutional neural network[J].Journal of Computer Applications, 2019, 39(12):3445-3449.(in Chinese)
[11] SINDAGI V A, PATEL V M.Generating high-quality crowd density maps using contextual pyramid CNNs[C]//Proceedings of IEEE International Conference on Computer Vision.Washington D.C., USA:IEEE Press, 2017:1879-1888.
[12] LI Y H, ZHANG X F, CHEN D M.CSRNet:dilated convolutional neural networks for understanding the highly congested scenes[C]//Proceedings of IEEE/CVF Conference on Computer Vision and Pattern Recognition.Washington D.C., USA:IEEE Press, 2018:1091-1100.
[13] SONG Q Y, WANG C G, WANG Y B, et al.To choose or to fuse?scale selection for crowd counting[C]//Proceedings of the 35th AAAI Conference on Artificial Intelligence.[S.l.]:AAAI Press, 2021:2576-2583.
[14] 李佳倩, 严华.基于跨列特征融合的人群计数方法[J].计算机科学, 2021, 48(6):118-124. LI J Q, YAN H.Crowd counting method based on cross-column features fusion[J].Computer Science, 2021, 48(6):118-124.(in Chinese)
[15] IDREES H, SALEEMI I, SEIBERT C, et al.Multi-source multi-scale counting in extremely dense crowd images[C]//Proceedings of IEEE Conference on Computer Vision and Pattern Recognition.Washington D.C., USA:IEEE Press, 2013:2547-2554.
[16] ZENG L K, XU X M, CAI B L, et al.Multi-scale convolutional neural networks for crowd counting[C]//Proceedings of IEEE International Conference on Image Processing.Washington D.C., USA:IEEE Press, 2017:465-469.
[17] KALYANI G, JANAKIRAMAIAH B, PRASAD L V N, et al.Efficient crowd counting model using feature pyramid network and ResNeXt[J].Soft Computing, 2021, 25(15):10497-10507.
[18] SHI M J, YANG Z H, XU C, et al.Revisiting perspective information for efficient crowd counting[C]//Proceedings of IEEE/CVF Conference on Computer Vision and Pattern Recognition.Washington D.C., USA:IEEE Press, 2019:7271-7280.
[19] ZHUGE J C, DING N N, XING S J, et al.An improved deep multiscale crowd counting network with perspective awareness[J].Optoelectronics Letters, 2021, 17(6):367-372.
[20] ZHANG Y M, ZHOU C L, CHANG F L, et al.Multi-resolution attention convolutional neural network for crowd counting[J].Neurocomputing, 2019, 329:144-152.
[21] CHEN J W, SU W, WANG Z F.Crowd counting with crowd attention convolutional neural network[J].Neurocomputing, 2020, 382:210-220.
[22] 翟强, 王陆洋, 殷保群, 等.基于尺度自适应卷积神经网络的人群计数算法[J].计算机工程, 2020, 46(2):250-254, 261. ZHAI Q, WANG L Y, YIN B Q, et al.Crowd counting algorithm based on scale adaptive convolutional neural network[J].Computer Engineering, 2020, 46(2):250-254, 261.(in Chinese)
[23] JIANG M Y, LIN J Z, WANG Z J.A smartly simple way for joint crowd counting and localization[J].Neurocomputing, 2021, 459:35-43.
[24] 袁健, 王姗姗, 罗英伟.基于图像视野划分的公共场所人群计数模型[J].计算机应用研究, 2021, 38(4):1256-1260, 1280. YUAN J, WANG S S, LUO Y W.Public place crowd counting model based on image field division[J].Application Research of Computers, 2021, 38(4):1256-1260, 1280.(in Chinese)
[25] WALACH E, WOLF L.Learning to count with CNN boosting[C]//Proceedings of European Conference on Computer Vision.Berlin:Springer, 2016:660-676.
[26] LIU L B, WANG H J, LI G B, et al.Crowd counting using deep recurrent spatial-aware network[C]//Proceedings of the 27th International Joint Conference on Artificial Intelligence.New York, USA:ACM Press, 2018:849-855.
[27] ZOU Z K, SHAO H L, QU X Y, et al.Enhanced 3D convolutional networks for crowd counting[EB/OL].[2021-09-22].https://arxiv.org/abs/1908.04121.
[28] LIU J, GAO C Q, MENG D Y, et al.DecideNet:counting varying density crowds through attention guided detection and density estimation[C]//Proceedings of IEEE/CVF Conference on Computer Vision and Pattern Recognition.Washington D.C., USA:IEEE Press, 2018:5197-5206.
[29] ZHOU J T, ZHANG L, DU J W, et al.Locality-aware crowd counting[J].IEEE Transactions on Pattern Analysis and Machine Intelligence, 2021, 65:99-103.
[30] HOSSAIN M, HOSSEINZADEH M, CHANDA O, et al.Crowd counting using scale-aware attention networks[C]//Proceedings of IEEE Winter Conference on Applications of Computer Vision.Washington D.C., USA:IEEE Press, 2019:1280-1288.

选择文件类型/文献管理软件名称

选择包含的内容