多尺度特征自适应融合的轻量化织物瑕疵检测

doi:10.19678/j.issn.1000-3428.0063507

摘要/Abstract

摘要： 织物瑕疵检测是纺织行业保证产品质量的重要环节，针对织物瑕疵检测中存在小目标瑕疵检测困难、不同种类瑕疵长宽比差异大、对实时性要求高等问题，提出一种新的轻量化织物瑕疵检测算法。以YOLOv4网络为基础，使用轻量化网络MobileNetv2为主干网络，有效减少模型参数总量与运算量，以满足实时性需求。在MobileNetv2的逆残差结构中加入CoordAttention注意力模块，将空间精确位置信息嵌入到通道注意力中，增强网络聚焦小目标特征的能力。使用自适应空间特征融合（ASFF）网络改进路径聚合网络（PANet），使模型通过学习获得多尺度特征图的融合权重，从而充分利用浅层特征与深层特征，提高算法对小目标瑕疵的检测精度。采用K-means++算法确定先验框尺寸，并用Focal Loss函数修改模型损失函数，降低正、负样本不平衡对检测结果的影响，解决不同种类瑕疵长宽比差异大及类别不平衡的问题。实验结果表明，相较于YOLOv4算法，所提算法的平均精度均值提高了2.3个百分点，检测速度提升了12 frame/s，能较好地应用于织物瑕疵检测。

关键词: 织物瑕疵检测, 自适应空间特征融合, CoordAttention模块, YOLOv4网络, MobileNetv2网络

Abstract: Fabric defect detection is an essential process in the textile industry for ensuring product quality.A algorithm for detecting lightweight fabric defects based on multiscale feature adaptive fusion is proposed to minimize small-target defect detection difficulties in fabric defect detection and the significant differences in the aspect ratios of different defects and high requirements for real-time performance.Based on the YOLOv4 network, lightweight network MobileNetv2 is used as the backbone network to effectively reduce the total number of model parameters and the cost of calculations to satisfy real-time requirements.In the inverse residual structure of MobileNetv2, a new attention mechanism is added.CoordAttention module, which embeds the spatially accurate position information necessary for detecting small-target defects into the channel attention, is used to enhance the ability of the network to focus on small-target features.Second, the Adaptive Spatial Feature Fusion(ASFF) network is used to improve Path Aggregation Network(PANet) to enable the model to obtain the fusion weights of multiscale feature maps through learning, fully utilize shallow and deep features, and further improve the detection accuracy of small-target defects.For different defect types and problems with significant aspect ratio differences and unbalanced categories, the K-means++ algorithm is used to determine the prior frame size, and the Focal Loss function is used to modify the model loss function to reduce the impact of the positive and negative sample imbalance on the detection result.The experimental results show that compared with YOLOv4, the mean Average Precision(mAP) of the proposed algorithm increases by 2.3 percentage points, and the detection speed increases by 12 frame/s.The proposed approach can be effectively applied to fabric defect detection.

Key words: fabric defect detection, Adaptively Spatial Feature Fusion(ASFF), CoordAttention module, YOLOv4 network, MobileNetv2 network

中图分类号:

TP393

杨毅, 桑庆兵. 多尺度特征自适应融合的轻量化织物瑕疵检测[J]. 计算机工程, 2022, 48(12): 288-295.

YANG Yi, SANG Qingbing. Lightweight-Fabric Defect Detection Based on Adaptive Fusion of Multiscale Features[J]. Computer Engineering, 2022, 48(12): 288-295.

http://www.ecice06.com/CN/Y2022/V48/I12/288

图/表 10

20230112184405

20230112184408

20230112184412

20230112184415

20230112184419

20230112184423

20230112184426

20230112184433

20230112184437

20230112184445

参考文献

[1] KUMAR A.Computer-vision-based fabric defect detection:a survey[J].IEEE Transactions on Industrial Electronics, 2008, 55(1):348-363.
[2] DING S M, LIU Z F, LI C L.AdaBoost learning for fabric defect detection based on HOG and SVM[C]//Proceedings of International Conference on Multimedia Technology.Washington D.C., USA:IEEE Press, 2011:2903-2906.
[3] HANBAY K, TALU M F, ÖZGÜVEN Ö F.Fabric defect detection systems and methods-a systematic literature review[J].Optik, 2016, 127(24):11960-11973.
[4] MIRMAHDAVI S A, AHMADYFARD A, SHAHRAKI A A, et al.A novel modeling of random textures using Fourier transform for defect detection[C]//Proceedings of the 15th International Conference on Computer Modelling and Simulation.Washington D.C., USA:IEEE Press, 2013:470-475.
[5] 任欢欢, 景军锋, 张缓缓, 等.应用GIS和FTDT的织物错花缺陷检测研究[J].激光与光电子学进展, 2019, 56(13):94-99. REN H H, JING J F, ZHANG H H, et al.Cross-printing defect detection of printed fabric using GIS and FTDT[J].Laser &Optoelectronics Progress, 2019, 56(13):94-99.(in Chinese)
[6] LIU L L, ZHANG H J, XU X F, et al.Collocating clothes with generative adversarial networks cosupervised by categories and attributes:a multidiscriminator framework[J].IEEE Transactions on Neural Networks and Learning Systems, 2020, 31(9):3540-3554.
[7] GOODFELLOW I, POUGET-ABADIE J, MIRZA M, et al.Conditional generative adversarial nets[J].Communications of the ACM, 2020, 63(11):139-144.
[8] ZHAO M B, LIU Y, LI X R, et al.An end-to-end framework for clothing collocation based on semantic feature fusion[J].IEEE MultiMedia, 2020, 27(4):122-132.
[9] SCHUSTER M, PALIWAL K K.Bidirectional recurrent neural networks[J].IEEE Transactions on Signal Processing, 1997, 45(11):2673-2681.
[10] ZHANG H J, WANG X H, LIU L L, et al.WarpClothingOut:a stepwise framework for clothes translation from the human body to tiled images[J].IEEE MultiMedia, 2020, 27(4):58-68.
[11] GOODFELLOW I, POUGET-ABADIE J, MIRZA M, et al.Generative adversarial nets[C]//Proceedings of Advances in Neural Information Processing Systems.New York, USA:ACM Press, 2014:2672-2680.
[12] 许玉格, 钟铭, 吴宗泽, 等.基于深度学习的纹理布匹瑕疵检测方法[J].自动化学报, 2020, 45(7):1-15. XU Y G, ZHONG M, WU Z Z, et al.Detection of detecting textured fabric defects based on deep learning.[J].Acta Automatica Sinica, 2020, 45(7):1-15.(in Chinese)
[13] HE K M, ZHANG X Y, REN S Q, et al.Deep residual learning for image recognition[C]//Proceedings of IEEE Conference on Computer Vision and Pattern Recognition.Washington D.C., USA:IEEE Press, 2016:770-778.
[14] BOCHKOVSKIY A, WANG C Y, LIAO H Y M.YOLOv4:optimal speed and accuracy of object detection[EB/OL].[2021-11-03].https://arxiv.org/abs/2004.10934.
[15] SANDLER M, HOWARD A, ZHU M L, et al.MobileNetV2:inverted residuals and linear bottlenecks[C]//Proceedings of IEEE/CVF Conference on Computer Vision and Pattern Recognition.Washington D.C., USA:IEEE Press, 2018:4510-4520.
[16] 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.
[17] LIU S T, HUANG D, WANG Y H.Learning spatial fusion for single-shot object detection[EB/OL].[2021-11-03].https://arxiv.org/abs/1911.09516.
[18] LIU S, QI L, QIN H F, et al.Path aggregation network for instance segmentation[C]//Proceedings of IEEE/CVF Conference on Computer Vision and Pattern Recognition.Washington D.C., USA:IEEE Press, 2018:8759-8768.
[19] ARTHUR D, VASSILVITSKII S.K-means++:the advantages of careful seeding[C]//Proceedings of the 18th Annual ACM-SIAM Symposium on Discrete Algorithms.New York, USA:ACM Press, 2007:1027-1035.
[20] 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.
[21] LIU W, ANGUELOV D, ERHAN D, et al.SSD:single shot multibox detector[C]//Proceedings of European Conference on Computer Vision.Berlin, Germany:Springer, 2016:21-37.
[22] REN S, HE K, 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.
[23] 吴涛, 王伟斌, 于力, 等.轻量级YOLOV3的绝缘子缺陷检测方法[J].计算机工程, 2019, 45(8):275-280. WU T, WANG W B, YU L, et al.Insulator defect detection method for lightweight YOLOV3[J].Computer Engineering, 2019, 45(8):275-280.(in Chinese)
[24] WANG C Y, MARK LIAO H Y, WU Y H, et al.CSPNet:a new backbone that can enhance learning capability of CNN[C]//Proceedings of IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops.Washington D.C., USA:IEEE Press, 2020:1571-1580.
[25] 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.
[26] 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.
[27] HU J, SHEN L, SUN G.Squeeze-and-excitation networks[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition.Washington D.C., USA:IEEE Press, 2018:7132-7141.
[28] WOO S, PARK J, LEE J Y, et al.CBAM:convolutional block attention module[C]//Proceedings of the European Conference on Computer Vision.Berlin, Germany:Springer, 2018:3-19.

选择文件类型/文献管理软件名称

选择包含的内容