[1]吕璐璐,陈树越,王利平,等.水体微纤维图像识别的改进MobileNetV2算法[J].郑州大学学报(工学版),2021,42(5):25-31.[doi:10.13705/j.issn.1671-6833.2021.05.005]
　Lu Lulu,Chen Shuyue,Wang Liping,et al.An Improved MobileNetV2 Algorithm for Image Recognition of Microfibers in Water[J].Journal of Zhengzhou University (Engineering Science),2021,42(5):25-31.[doi:10.13705/j.issn.1671-6833.2021.05.005]
点击复制

水体微纤维图像识别的改进MobileNetV2算法()
分享到：

《郑州大学学报(工学版)》[ISSN:1671-6833/CN:41-1339/T]

卷:
42
期数:
2021年5期
页码:
25-31
栏目:
出版日期:
2021-09-10

文章信息/Info

Title:
An Improved MobileNetV2 Algorithm for Image Recognition of Microfibers in Water
作者:
吕璐璐1,陈树越1,王利平2,许霞2
1.常州大学 微电子与控制工程学院，江苏 常州 213164；2.常州大学 环境与安全工程学院，江苏 常州 213164
Author(s):
Lu Lulu1; Chen Shuyue1; Wang Liping2; Xu Xia2;
1.School of Microelectronic and Control Engineering， Changzhou University， Changzhou 213164， China； 2.School of Environmental &Safety Engineering， Changzhou University， Changzhou 213164， China
关键词:
Keywords:
waterbody;  microfiber recognition;  MobileNetV2;  pooling fusion;  feature reconstruction
DOI:
10.13705/j.issn.1671-6833.2021.05.005
文献标志码:
A
摘要:
针对人工识别水体微纤维耗时耗力,以及传统图像处理算法识别水体微纤维图像鲁棒性弱等问题,构建了一种改进的 MobileNetV2 网络识别微纤维算法。 在特征提取部分采用特征重构策略,先压缩深度卷积特征,获取全局感受野；再利用多层全连接为每个通道生成权重,建立通道之间的相互依赖关系；最后逐通道加权到原特征上,完成对原始特征的重构。 此外,采用不同大小的下采样器捕获不同尺度的特征信息并融合,增强微纤维的细节特征信息,提升模型对微纤维的学习能力与识别效果。 改进MobileNetV2 网络的微纤维识别准确率达到 97.96%,与原始 MobileNetV2 网络相比高 2.54%,同时,误识率和漏识率也有显著的降低。 相较于 ResNet、DenseNet、VGG16 和 NasNet 网络,模型大小压缩 了若干倍,微纤维识别准确率有所提升,误识率与漏识率大大降低。 实验表明:该网络模型能够提取更加完整的微纤维特征信息,加强微纤维特征判别指向性的同时减小了模型尺寸,降低了在移动设备中部署的难度,并且使识别微纤维具有更高的准确率和更好的稳定性。
Abstract:
Aiming at the problems of time-consuming and labor-consuming manual identification of water microfibers, and the weak robustness of traditional image processing algorithms for identifying water microfiber images, an improved MobileNetV2 network identification method for microfibers is constructed. In the feature extraction part, the feature reconstruction strategy is adopted. Firstly, the deep convolution features are compressed to obtain the global receptive field. Then, the fully connected layers are used to generate weights for each channel to establish the interdependence between the channels. Finally, the channel is weighted to the original in terms of features to complete the reconstruction of the original features. In addition, different sizes of downsamplers are used to capture and fuse feature information of different scales to enhance the detailed feature information of microfibers, and to improve the model′s learning ability and recognition effect of microfibers. The improved MobileNetV2 network′s microfiber recognition accuracy rate reaches 97.96%. Compared with the original MobileNetV2 network, the recognition accuracy rate is increased by 2.54%. At the same time, the false recognition rate and the missed recognition rate are also significantly reduced. In comparison to ResNet, DenseNet, VGG16 and NasNet networks, the model size is compressed several times, the accuracy of microfiber recognition is improved, and the false recognition rate and missed recognition rate are greatly reduced. Experimental results show that the network model can extract more complete feature information for microfiber. While strengthening the microfiber feature to identify the directivity, the model is reduced, and the difficulty of deployment in mobile devices is reduced as well. The improved model recognizes microfibers with higher accuracy and better stability.

参考文献/References:

[1] THOMPSON R C，OLSEN Y，MITCHELL R P, et al.Lost at sea： where is all the plastic？ [J].Science，2004，304（5672） ：838.

[2] HODSON M E ，DUFFUS-HODSON C A，CLARK A，et al.Plastic bag derived-microplastics as a vector for metal exposure in terrestrial invertebrates[J].Environmental science & technology， 2017， 51 （ 8 ） ： 4714-4721.
[3] ANDRADY A L.Microplastics in the marine environment [J].Marine pollution bulletin， 2011， 62（8） ：1596-1605.
[4] SMITH M，LOVE D C，ROCHMAN C M，et al.Microplastics in seafood and the implic ations for human health[ J].Current environmental health reports，2018，5（3） ：375-386.[5] 许霞，侯青桐，薛银刚，等.污水厂中微塑料的污染及迁移特征研究进展 [J].中国环境科学，2018，38（11） ：4393-4400.
[6] MILLER R Z，WATTS A J R，WINSLOW B O，et al. Mountains to the sea： river study of plastic and nonplastic microfiber pollution in the northeast USA [J]. Marine pollution bulletin，2017，124（1） ：245-251.
[7] 李珊，张岚，陈永艳，等.饮用水中微塑料检测技术研究进展[J].净水技术，2019，38（4） ：1-8.
[8] 李剑平.扫描电子显微镜对样品的要求及样品的制备[J].分析测试技术与仪器，2007，13（1） ：74-77.
[9] SILVA A B， BASTOS A S， JUSTINO C I L， et al. Microplastics in the environment： challenges in analytic al chemistry-a review [J].Analytic a chimic aacta，2018，1017：1-19.
[10] YURTSEVER M， YURTSEVER U. Use of a convolutional neural network for the classific ation of microbeads in urban wastewater [J].Chemosphere，2019，216：271-280.
[11] PAZDERNIK K，LAHAYE N L，ARTMAN C M，et al. Microstructural classific ation of unirradiated LiAlO2 pellets by deep learning methods [J].Computational materials science，2020，181：109728.
[12] 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.
[13] HE K M，ZHANG X Y，REN S Q，et al.Spatial pyramid pooling in deep convolutional networks for visual recognition[J].IEEE transactions on pattern analysis and machine intelligence，2015，37（9） ：1904-1916.
[14] HE K M，ZHANG X Y，REN S Q，et al.Deep residual learning for image recognition [C] //2016 IEEE Conference on Computer Vision and Pattern Recognition（CVPR）.Pisc ataway：IEEE ，2016：770-778.
[15] HUANG G， LIU Z， VAN DER MAATEN L，et al.Densely connected convolutional networks [C] //2017 IEEE Conference on Computer Vision and Pattern Recognition（CVPR）.Pisc ataway：IEEE ， 2017： 2261-2269.
[16] SIMONYAN K， ZISSERMAN A.Very deep convolutional networks for large-sc ale image recognition[EB/OL].（2014-09-04） [2020-10-30].https：//arxiv.org/abs/1409.1556.
[17] ZOPH B，VASUDEVAN V，SHLENS J，et al.Learning transferable architectures for sc alable image recognition[C] //2018 IEEE /CVF Conference on Computer Vision and Pattern Recognition.Pisc ataway： IEEE ，2018：8697-8710.
[18] SANDLER M，HOWARD A， ZHU M L， et al.MobileNetV2：inverted residuals and linear bottlenecks[C] //2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition.Pisc ataway： IEEE ， 2018： 4510-4520.

更新日期/Last Update: 2021-10-11