基于改进SSD的口罩佩戴检测算法

doi:10.19678/j.issn.1000-3428.0062150

摘要/Abstract

摘要： 新冠疫情近年来在全球肆虐，新冠病毒具有极强的传染性，在公共场所佩戴口罩可以阻断病毒的传播途径，有效遏止疫情的蔓延。利用计算机视觉技术对公共场合佩戴口罩行为进行检测具有重要意义，在疫情防控常态化条件下，需要对人脸口罩佩戴进行正确识别，同时要识别口罩佩戴是否正确。在实际检测环境中，口罩佩戴检测任务中的场景复杂多样，佩戴口罩的人脸目标尺度不一，正确与错误佩戴口罩特征差异小，难以检测。提出一种改进SSD算法的口罩佩戴检测算法。以SSD检测算法为基础，引入特征融合网络及协调注意力机制，重构特征提取网络，增强对细节信息的学习和处理能力。同时，将算法的分类预测分数和IoU分数进行合并表示，使用Quality Focal Loss函数调节正负样本的权重。在自制口罩佩戴检测数据集上的实验结果表明，该算法的平均精度均值达到96.28%，与原始算法相比提高了5.62%，对口罩佩戴检测具有良好的准确性和实用性，可满足疫情防控下的实际需求。

关键词: 口罩佩戴检测, SDD检测算法, 特征融合, 注意力机制, 损失函数

Abstract: In recent years, the COVID-19, which involves a highly infectious virus, has spread worldwide.Wearing masks in public areas can reduce the transmission and hence the spread of the virus.Additionally, using computer vision technology to detect mask wearing behavior in public areas is crucial.To prevent and control epidemics, the correct form of wearing face masks must be identified.In an actual environment, the detection of mask wearing is complex and diverse.The scale of a face wearing a mask is different;furthermore, the difference between the correct and wrong forms of wearing a mask is subtle and hence difficult to detect.Therefore, a mask wearing detection algorithm based on an improved Single Shot Multibox Detector(SSD) algorithm is proposed herein.Based on the SSD network, the algorithm introduces a feature fusion network and an attention coordination mechanism, reconstructs the feature extraction network, and enhances the ability of learning and processing detailed information.In addition, the classification prediction score and IoU score of the algorithm are combined, whereas the Quality Focal Loss(QFL) function is used to adjust the weight of positive and negative samples.An experiment is performed on acustom-developed mask wearing test dataset.Experimental results show that the average accuracy of the algorithm is 96.28%, which is 5.62% higher than that of the original algorithm.The improved algorithm offers good accuracy and practicability for mask wearing detection, as well assatisfies the requirements for epidemic prevention and control.

Key words: mask wearing detection, SDD detection algorithm, feature fusion, attention mechanism, loss function

中图分类号:

TP391.4

李雨阳, 沈记全, 翟海霞, 冯伟华. 基于改进SSD的口罩佩戴检测算法[J]. 计算机工程, 2022, 48(8): 173-179,186.

LI Yuyang, SHEN Jiquan, ZHAI Haixia, FENG Weihua. Mask Wearing Detection Algorithm Based on Improved SSD[J]. Computer Engineering, 2022, 48(8): 173-179,186.

http://www.ecice06.com/CN/Y2022/V48/I8/173

图/表 16

20220825092353

20220825092357

20220825092406

20220825092410

20220825092414

20220825092425

20220825092429

20220825092434

20220825092438

20220825092443

20220825092447

20220825092450

20220825092457

20220825092501

20220825092505

20220825092509

参考文献

[1] LEUNG N H L, CHU D K W, SHIU E Y C, et al.Respiratory virus shedding in exhaled breath and efficacy of face masks[J].Nature Medicine, 2020, 26(5):676-680.
[2] HE Y M, WANG Z H, GUO S Y, et al.Face mask detection algorithm based on HSV+HOG features and SVM[EB/OL].[2021-06-10].https://kns.cnki.net/kcms/detail/14.1357.TH.20210315.0832.002.html.
[3] JIANG M, FAN X, YAN H.Retina facemask:a face mask detector[EB/OL].[2021-06-10].https://arxiv preprint arxiv:2005.03950.
[4] ABBASI S, ABDI H, AHMADI A.A face-mask detection approach based on YOLO applied for a new collected dataset[C]//Proceedings of the 26th International Conference on Computer Vision and Pattern Recognition.Washington D.C., USA:IEEE Press, 2021:1-6.
[5] 李文斌, 何冉.基于深度神经网络的遥感图像飞机目标检测[J].计算机工程, 2020, 46(7):268-276. LI W B, HE R.Aircraft target detection of remote sensing images based on deep neural network[J].Computer Engineering, 2020, 46(7):268-276.(in Chinese)
[6] LIU G X, NOUAZE J C, TOUKO MBOUEMBE P L, et al.YOLO-tomato:a robust algorithm for tomato detection based on YOLOv3[J].Sensors, 2020, 20(7):2145.
[7] 程淑红, 周斌.基于改进CNN的铝轮毂背腔字符识别[J].计算机工程, 2019, 45(5):182-186. CHENG S H, ZHOU B.Recognition of characters in aluminum wheel back cavity based on improved convolution neural network[J].Computer Engineering, 2019, 45(5):182-186.(in Chinese)
[8] GIRSHICK R.Fast R-CNN[C]//Proceedings of 2015 IEEE International Conference on Computer Vision.Washington D.C., USA:IEEE Press, 2015:1440-1448.
[9] REN S Q, HE K M, GIRSHICK R, et al.Faster R-CNN:towards real-time object detection with region proposal networks[J].IEEE Transactions on Pattern Analysis and Machine Intelligence, 2017, 39(6):1137-1149.
[10] REDMON J, DIVVALA S, GIRSHICK R, et al.You only look once:unified, real-time object detection[C]//Proceedings of 2016 IEEE Conference on Computer Vision and Pattern Recognition.Washington D.C., USA:IEEE Press, 2016:779-788.
[11] REDMON J, FARHADI A.YOLO9000:better, faster, stronger[C]//Proceedings of 2017 IEEE Conference on Computer Vision and Pattern Recognition.Washington D.C., USA:IEEE Press, 2017:6517-6525.
[12] REDMON J, FARHADI A.YOLOv3:an incremental improvement[EB/OL].[2021-06-10].https://arxiv.org/abs/1804.02767.
[13] LIU W, ANGUELOVD, ERHAN D, et al.SSD:single shot MultiBox detector[C]//Proceedings of 2016 European Conference on Computer Vision.Berlin, Germany:Springer, 2016:21-37.
[14] MORERA Á, SÁNCHEZ Á, MORENO A B, et al.SSD vs.YOLO for detection of outdoor urban advertising panels under multiple variabilities[J].Sensors, 2020, 20(16):4587.
[15] WANG Y B, WANG Y J, DANG L F.Video detection of foreign objects on the surface of belt conveyor underground coal mine based on improved SSD[J].Journal of Ambient Intelligence and Humanized Computing, 2020, 56:1-10.
[16] FANG J Z, YANG H Y, CHEN P, et al.A detection algorithm of giant panda in wild video image based on wavelet-SSD network[C]//Proceedings of IEEE International Conference on Systems, Man, and Cybernetics.Washington D.C., USA:IEEE Press, 2020:3655-3660.
[17] 张震, 李孟洲, 李浩方, 等.改进SSD算法及其在地铁安检中的应用[J].计算机工程, 2021, 47(7):314-320. ZHANG Z, LI M Z, LI H F, et al.Improved SSD algorithm and its application in subway security detection[J].Computer Engineering, 2021, 47(7):314-320.(in Chinese)
[18] LIN T Y, DOLLÁR P, GIRSHICK R, et al.Feature pyramid networks for object detection[C]//Proceedings of IEEE Conference on Computer Vision and Pattern Recognition.Washington D.C., USA:IEEE Press, 2017:936-944.
[19] HOU Q B, ZHOU D Q, FENG J S.Coordinate attention for efficient mobile network design[C]//Proceedings of IEEE/CVF Conference on Computer Vision and Pattern Recognition.Washington D.C., USA:IEEE Press, 2021:13708-13717.
[20] SIMONYAN K, ZISSERMAN A.Very deep convolutional networks for large-scale image recognition[EB/OL].[2021-06-10].https://arxiv.org/abs/1409.1556.
[21] LIN T Y, GOYAL P, GIRSHICK R, et al.Focal loss for dense object detection[C]//Proceedings of IEEE International Conference on Computer Vision.Washington D.C., USA:IEEE Press, 2017:2999-3007.
[22] LI X, WANG W H, WU L J, et al.Generalized focal loss:learning qualified and distributed bounding boxes for dense object detection[EB/OL].[2021-06-10].https://arxiv preprint arxiv:2006.04388.

选择文件类型/文献管理软件名称

选择包含的内容