|本期目录/Table of Contents|

[1]陈江川,吴云韬*,孔 权.基于CBAM-Res2Net的人群计数算法[J].武汉工程大学学报,2022,44(06):664-669.[doi:10.19843/j.cnki.CN42-1779/TQ.202206042]
 CHEN Jiangchuan,WU Yuntao*,KONG Quan.Crowd Counting Algorithm Based on CBAM-Res2Net[J].Journal of Wuhan Institute of Technology,2022,44(06):664-669.[doi:10.19843/j.cnki.CN42-1779/TQ.202206042]
点击复制

基于CBAM-Res2Net的人群计数算法(/HTML)
分享到:

《武汉工程大学学报》[ISSN:1674-2869/CN:42-1779/TQ]

卷:
44
期数:
2022年06期
页码:
664-669
栏目:
机电与信息工程
出版日期:
2022-12-31

文章信息/Info

Title:
Crowd Counting Algorithm Based on CBAM-Res2Net
文章编号:
1674 - 2869(2022)06 - 0664 - 06
作者:
陈江川12吴云韬*12孔 权3
1. 武汉工程大学计算机科学与工程学院,湖北 武汉 430205;
2. 智能机器人湖北省重点实验室(武汉工程大学),湖北 武汉 430205;
3. 武汉工程大学艺术设计学院,湖北 武汉 430205
Author(s):
CHEN Jiangchuan12 WU Yuntao*12 KONG Quan3
1. School of Computer Science & Engineering, Wuhan Institute of Technology, Wuhan 430205, China;
2. Hubei Key Laboratory of Intelligent Robot(Wuhan Institute of Technology), Wuhan 430205, China;
3. School of Art & Design, Wuhan Institute of Technology, Wuhan 430205, China

关键词:
人群计数多尺度特征提取模块卷积注意力模块CBAM-Res2Net密度图
Keywords:
crowd countingmulti-scale feature extraction moduleconvolutional block attention moduleCBAM-Res2Netdensity map
分类号:
TP391
DOI:
10.19843/j.cnki.CN42-1779/TQ.202206042
文献标志码:
A
摘要:
针对静态人群图像中背景干扰和尺度变化等问题,采用多尺度特征提取模块(Res2Net)以更细的粒度提取多尺度特征,提高对不同尺寸人头的计数性能;引入卷积注意力模块(CBAM),分别在通道域和空间域上提高人群区域的权重,有效改善了高密度和复杂的人群场景下背景干扰等问题。在此基础上,将CBAM模块集成到Res2Net模块中,形成了新的多尺度特征提取模块CBAM-Res2Net。在后端网络中设计了一个扩张模块以提取更深层的特征并进行特征融合回归,从而生成高质量的密度图。并且分别在ShanghaiTech Part A、ShanghaiTech Part B和UCF_CC_50数据集上进行了算法对比实验,本文模型在上述数据集的平均绝对误差和均方根误差分别为61.4、7.3、255.6和98.5、10.8、310.2,综合性能均优于其他算法,验证了模型的准确性和鲁棒性。
Abstract:
Aimed at the problems of background interference and scale change in static crowd images, the multi-scale feature extraction module (Res2Net) was used to extract multi-scale features with finer granularity to improve the counting performance of heads with different sizes. The convolutional block attention module (CBAM) was introduced to improve the weight of the crowd area in the channel domain and spatial domain respectively, effectively improving background interference and other problems in high-density and complex crowd scenes. On this basis, the CBAM module was integrated into Res2Net module, and a new multi-scale feature extraction module CBAM-Res2Net was proposed. An expansion module was designed in the back-end network to extract better features and perform feature fusion regression for generating high-quality density maps. The algorithm comparison experiments were carried out on ShanghaiTech Part A, ShanghaiTech Part B and UCF_CC_50 datasets respectively. The mean absolute errors and the root mean square errors of the proposed model in the above datasets are 61.4, 7.3, 255.6 and 98.5, 10.8, 310.2, respectively. The comprehensive performances of the proposed model are better than those of other algorithms, which verifies the accuracy and robustness of the model.

参考文献/References:

[1] SHAO J, KANG K, LOY C C, et al. Deeply learned attributes for crowded scene understanding[C]// 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).Boston:IEEE,2015:4657-4666.

[2] CHEN K, GONG S G, XIANG T, et al. Cumulative attribute space for age and crowd density estimation[C]// 2013 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Portland: IEEE, 2013: 2467-2474.
[3] 田月媛,邓淼磊,高辉,等.基于深度学习的人群计数算法综述[J].电子测量技术,2022,45(7):152-159.
[4] 卢振坤,刘胜,钟乐,等.人群计数研究综述[J].计算机工程与应用, 2022,58(11):33-46.
[5] 余鹰,朱慧琳,钱进,等.基于深度学习的人群计数研究综述[J].计算机研究与发展,2021,58(12):2724-2747.
[6] ZHANG C, LI H S, WANG X G, et al. Cross-scene crowd counting via deep convolutional neural networks[C]// IEEE Conference on Computer Vision & Pattern Recognition (CVPR). Boston: IEEE, 2015: 833-841.
[7] ZHANG Y Y, ZHOU D S, CHEN S Q, et al. Single-Image crowd counting via multi-column convolutional neural network[C]// 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Las Vegas: IEEE, 2016: 589-597.
[8] SHI M J, YANG Z H, XU C, et al. Revisiting perspective information for efficient crowd counting[C]// 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Long Beach: IEEE, 2019: 7271-7280.
[9] LI Y H, ZHANG X F, CHEN D M. CSRNet: dilated convolutional neural networks for understanding the highly congested scenes[C]// 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Salt Lake City: IEEE, 2018: 1091-1100.
[10] LIU N, LONG Y C, ZOU C Q, et al. ADCrowdNet: an attention-injective deformable convolutional network for crowd understanding[C]// 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Long Beach: IEEE, 2019: 3220-3229.
[11] WOO S, PARK J, LEE J Y, et al. Cbam: Convolutional block attention module[C]//Proceedings of the European conference on computer vision (ECCV). Berlin: Springer, 2018: 3-19.
[12] GAO S H, CHENG M M, ZHAO K, et al. Res2Net: A new multi-scale backbone architecture[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2021,43(2): 652-662.
[13] SIMONYAN K, ZISSERMAN A. Very deep convolutional networks for large-scale image recognition[C]// International Conference on Learning Representations (ICLR). San Diego: OpenReview.net, 2015: 1-14.
[14] HU J, SHEN L, ALBANIE S, et al. Squeeze-and-excitation networks[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2020,42(8): 2011-2023.
[15] SINDAGI V A, PATEL V M. Generating high-quality crowd density maps using contextual pyramid CNNs[C]// 2017 IEEE International Conference on Computer Vision (ICCV). Venice: IEEE, 2017: 1879-1888.
[16] SHI M J, YANG Z H, XU C, et al. Revisiting perspective information for efficient crowd counting[C]// 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Long Beach: IEEE, 2019: 7271-7280.
[17] WU X J, ZHENG Y B, YE H, et al. Adaptive scenario discovery for crowd counting[C]// ICASSP 2019-2019 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). Brighton: IEEE, 2019: 2382-2386.
[18] LIU W Z, SALZMANN M, FUA P. Context-aware crowd counting[C]// 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Long Beach: IEEE, 2019: 5094-5103.
[19] 吴奇元,王晓东,章联军,等.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.
[20] SAM D B, SURYA S, BABU R V. Switching convolutional neural network for crowd counting[C]// 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Honolulu: IEEE, 2017: 4031-4039.

相似文献/References:

备注/Memo

备注/Memo:
收稿日期:2022-06-28
基金项目:国家自然科学基金(61771353);湖北三峡实验室开放基金(SC215001)
作者简介:陈江川,硕士研究生。E-mail:2649359332@qq.com
*通讯作者:吴云韬,博士,教授。E-mail:ytwu@wit.edu.cn
引文格式:陈江川,吴云韬,孔权. 基于CBAM-Res2Net的人群计数算法[J]. 武汉工程大学学报,2022,44(6):664-669.

更新日期/Last Update: 2023-01-09