|本期目录/Table of Contents|

[1]徐梓涵,刘 军*,张苏沛,等.一种基于MobileNet的火灾烟雾检测方法[J].武汉工程大学学报,2019,(06):580-585.[doi:10. 3969/j. issn. 1674-2869. 2019. 06. 012]
 XU Zihan,LIUJun*,ZHANG Supei,et al.Fire and Smoke Detection Method Based on MobileNet[J].Journal of Wuhan Institute of Technology,2019,(06):580-585.[doi:10. 3969/j. issn. 1674-2869. 2019. 06. 012]
点击复制

一种基于MobileNet的火灾烟雾检测方法(/HTML)
分享到:

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

卷:
期数:
2019年06期
页码:
580-585
栏目:
机电与信息工程
出版日期:
2021-01-24

文章信息/Info

Title:
Fire and Smoke Detection Method Based on MobileNet
文章编号:
20190612
作者:
徐梓涵12刘 军*12张苏沛12肖澳文12杜 壮12
1. 智能机器人湖北省重点实验室(武汉工程大学),湖北 武汉 430205; 2. 武汉工程大学计算机科学与工程学院,湖北 武汉 430205
Author(s):
XU Zihan12 LIUJun*12 ZHANG Supei12 XIAO Aowen12 DU Zhuang12
1. Hubei Key Laboratory of Intelligent Robot(Wuhan Institute of Technology), Wuhan 430205, China; 2. School of Computer Science & Engineering, Wuhan Institute of Technology, Wuhan 430205, China;
关键词:
火灾烟雾检测MobileNetK-近邻算法
Keywords:
fire and smokedetection MobileNet K-nearest neighbor
分类号:
TP317.4
DOI:
10. 3969/j. issn. 1674-2869. 2019. 06. 012
文献标志码:
A
摘要:
提出了基于图像序列的火灾烟雾检测方法。首先使用K-近邻(K-NN)背景减除器预测前景区域,对该区域进行形态学操作后得到可能出现火焰或烟雾的区域。其次,使用轻量神经网络MobileNet对火焰和烟雾进行分类。该模型具有流线型架构,同时采用depthwise separate convolution,使得该模型可以运行在嵌入式设备和普通PC机上。实验首先在数据集上完成分类模型训练,使用多种标准进行评估。结果表明:该方法能够在嵌入式设备等计算能力有限的设备上实现火灾烟雾检测。与其他模型相比,该方法在没有明显损失准确度的情况下大幅提高了检测效率。
Abstract:
The detection method based on image sequences was proposed in the paper. Firstly, the foreground area was extracted by a K-nearest neighbor classifier. Then, the potential area of fire and smoke was recognized based on morphological operations. Finally, a lightweight neural network MobileNet was used to classify fire and smoke. MobileNet has a streamlined architecture and employs the depth wise separate convolution, which makes it possible to run on both personal computer and embedded devices. In experiments, a classifier was trained on a dataset, and was evaluated according to multiple metrics. The results show that the proposed method is able to detect fire and smoke on embedded devices, and improves the detection efficiency without significant loss of accuracy in comparison with other methods.

参考文献/References:

[1] 杨帆, 陈茂林, 吴迅,等. 基于传感器信息融合技术的森林火灾报警系统[J]. 华中科技大学学报(自然科学版), 2013, 41(2): 22-25. [2] 方帅, 祁林娟, 于磊. 多特征分析的视频烟雾检测方法[J]. 计算机工程与应用, 2016, 52(13): 222-227. [3] 叶剑锋, 王化明. AdaBoost检测结合SOM的自动人脸识别方法[J]. 哈尔滨工程大学学报, 2018, 39(1): 129-134. [4] 张苏沛, 刘军, 肖澳文,等. 基于卷积神经网络的验证码识别[J]. 武汉工程大学学报, 2019, 41(1): 89-92. [5] SIMONYAN K, ZISSERMAN A. Very deep convolutional networks for large-scale image recognition[J]. arXiv preprint arXiv:1409.1556, 2014. [6] 姚太伟, 王慧琴, 胡燕. 基于小波变换和稀疏光流法的火灾烟雾检测[J]. 计算机工程, 2012, 38(6): 204-206. [7] 刘颖, 顾小东, 李大湘. 基于LSA和SVM的火灾烟雾检测算法[J]. 西安邮电大学学报, 2014, 19(6): 6-10. [8] MAO W T, WANG W P, DOU Z, et al. Fire recognition based on multi-channel convolutional neural network[J]. Fire technology, 2018, 54(2): 531-554. [9] 陈俊周, 汪子杰, 陈洪瀚,等. 基于级联卷积神经网络的视频动态烟雾检测[J]. 电子科技大学学报, 2016, 45(6): 992-996. [10] YIN Z J, WAN B Y, YUAN F N, et al. A deep normalization and convolutional neural network for image smoke detection[C]//Proceedings of the IEEE conference on compute vision and patten recognition. Hawaii:IEEE,2017(5): 18429-18438. [11] MUHAMMAD K, A HMAD J, MEHMOOD I ,et al. Convolutional neural networks based fire detection in surveillance videos[C]//Proceedings of the IEEE conference on compute vision and patten recognition. Kansas city:IEEE, 2018(6): 18174-18183. [12] 龙群芳, 张东. 一种基于 KNN 纹理分类的超声图像自动分割方法[J]. 科技通报, 2016, 32(7): 138-142. [13] 牛冲. 基于图像处理的草莓病害识别方法研究[D]. 太原: 太原理工大学, 2016. [14] ZIVKOVIC Z, VAN DER HEIJDEN F. Efficient adaptive density estimation per image pixel for the task of background subtraction[J]. Pattern Recognition Letters, 2006, 27(7): 773-780. [15] 赵亮, 骆炎民, 骆翔宇. 基于背景动态更新与暗通道先验的火灾烟雾检测算法[J]. 计算机应用研究, 2017, 34(3): 957-960. [16] 傅天驹, 郑嫦娥, 田野,等. 复杂背景下基于深度卷积神经网络的森林火灾识别[J]. 计算机与现代化, 2016 (3): 52-57. [17] XU X D, ZHANG H X, DAI X, et al. SDN based next generation mobile network with service slicing and trials[J]. China Communications, IEEE, 2014, 11(2): 65-77. [18] 王正来, 黄敏, 朱启兵,等. 基于并行深度残差网络的堆场烟雾检测方法[J]. 激光与光电子学进展, 2018, 55(5): 146-152.

相似文献/References:

[1]董方武,王绍卜,马子余.基于ZigBee的碱液质量分数在线检测系统设计[J].武汉工程大学学报,2010,(01):100.
 DONG Fang wu,WANG Shao bu,MA Zi yu.Design of realtime lye Mass fraction detection system based on ZigBee technology[J].Journal of Wuhan Institute of Technology,2010,(06):100.
[2]黎俊波,李楠楠,余响林*,等.高选择性铁离子荧光探针的合成及性质研究[J].武汉工程大学学报,2010,(05):11.[doi:10.3969/j.issn.16742869.2010.05.003]
 LI Jun bo,LI Nan nan,YU Xiang lin,et al.Synthesis and properties of the Fe3+ fluorescence probewith high selectivity[J].Journal of Wuhan Institute of Technology,2010,(06):11.[doi:10.3969/j.issn.16742869.2010.05.003]
[3]孙 威,熊双玉,彭掌珠,等.荧光多孔二氧化硅纳米微球对Hg+离子的选择性检测和去除[J].武汉工程大学学报,2020,42(05):501.[doi:10.19843/j.cnki.CN42-1779/TQ.202001012]
 SUN Wei,XIONG Shuangyu,PENG Zhangzhu,et al.Fluorescent Porous Silica Nanospheres for Selective Detection and Removal of Hg+ Ion[J].Journal of Wuhan Institute of Technology,2020,42(06):501.[doi:10.19843/j.cnki.CN42-1779/TQ.202001012]
[4]王永靖,余良浪,秦 琴*,等.室内空气甲醛检测及去除技术的最新进展[J].武汉工程大学学报,2022,44(05):482.[doi:10.19843/j.cnki.CN42-1779/TQ.202205010]
 WANG Yongjing,YU Lianglang,QIN Qin*,et al.Recent Advances of Indoor Air Formaldehyde Detection and?Removal Technologies[J].Journal of Wuhan Institute of Technology,2022,44(06):482.[doi:10.19843/j.cnki.CN42-1779/TQ.202205010]
[5]南齐钰,熊知萌,梁文杰,等.商业化试剂构建荧光探针用于肼的检测及防伪识别[J].武汉工程大学学报,2023,45(05):490.[doi:10.19843/j.cnki.CN42-1779/TQ.202210033]
 NAN Qiyu,XIONG Zhimeng,LIANG Wenjie,et al.Commercial Reagent as Fluorescent Probe for Hydrazine Detection and Anti-Counterfeiting Identification[J].Journal of Wuhan Institute of Technology,2023,45(06):490.[doi:10.19843/j.cnki.CN42-1779/TQ.202210033]

备注/Memo

备注/Memo:
收稿日期:2019-05-19 基金项目:国家自然科学基金(61172150,61803286);智能机器人湖北省重点实验室开放基金(HBIR 201802);武汉工程大学第十届研究生教育创新基金(CX2018197,CX2018200,CX2018212) 作者简介:徐梓涵。E-mail:[email protected] *通讯作者:刘 军,博士,副教授。E-mail:[email protected] 引文格式:徐梓涵,刘军,张苏沛,等. 一种基于MobileNet的火灾烟雾检测方法[J]. 武汉工程大学学报,2019,41(6):580-585.
更新日期/Last Update: 2020-01-20