[1]刘 正,吴诗豪,侯 珏,等.基于数据增强和改进卷积神经网络的织物纬斜检测[J].服装学报,2023,8(05):391-399.
　LIU Zheng,WU Shihao,HOU Jue,et al.Fabric Skewing Detection Based on Data Augmentation and Improved Convolutional Neural Network[J].Journal of Clothing Research,2023,8(05):391-399.
点击复制

基于数据增强和改进卷积神经网络的织物纬斜检测()
分享到：

《服装学报》[ISSN:2096-1928/CN:32-1864/TS]

卷:
第8卷
期数:
2023年05期
页码:
391-399
栏目:
服装智造
出版日期:
2023-10-31

文章信息/Info

Title:
Fabric Skewing Detection Based on Data Augmentation and Improved Convolutional Neural Network
作者:
刘 正1; 2;  吴诗豪1; 3;  侯 珏1; 3; 4;  杨 阳1; 3
1.浙江理工大学 浙江省服装工程技术研究中心,浙江 杭州 310018; 2.浙江理工大学 国际时装技术学院,浙江 杭州 310018; 3.浙江理工大学 服装学院,浙江 杭州 310018; 4.武汉纺织大学 武汉纺织服装数字化工程技术研究中心,湖北 武汉 430073
Author(s):
LIU Zheng1; 2;  WU Shihao1; 3;  HOU Jue1; 3; 4;  YANG Yang1; 3
1.Zhejiang Provincial Research Center of Clothing Engineering Technology,Zhejiang Sci-Tech University,Hangzhou 310018,China; 2.International Fashion Technology College,Zhejiang Sci-Tech University,Hangzhou 310018,China; 3.School of Fashion Design and Engineering,Zhejiang Sci-Tech University,Hangzhou 310018,China; 4.Wuhan Textile and Apparel Digital Engineering Technology Research Center,Wuhan Textile University,Wuhan 430073,China
分类号:
TS 105.1.12
文献标志码:
A
摘要:
纬斜是织物生产和后整理加工中常见的疵点,光电设备检测纬斜的方法效率低且不精确。为了提升纬斜疵点的检测效率,将神经网络运用到纬斜检测中,结合纬斜特征改进卷积神经网络,提出具有循环训练策略的目标识别网络——纬斜检测网络,并在网络中加入正样本回归和多尺度输入,以提升卷积网络的性能。为了获得充足的纬斜样本数据用于网络训练,提出一种纬斜疵点数据增强方法,通过将纬斜图像公式化生成大量纬斜样本,并采用综合比较实验评估纬斜检测网络性能。结果表明,纬斜检测网络在纬斜检测中表现出色,检测精度达到98%,平均F-score达到0.97,同时使纬斜率的误差控制在±8%以内,检测性能优于其他目标检测模型。与YOLO网络相比,纬斜检测网络在真实纬斜样本检测中性能优异,拥有良好的跨数据集检测性能。
Abstract:
Skewing is a common defect in fabric production and finishing processing. The method of detecting skewing by photoelectric equipment is inefficient and imprecise. To improve the detection efficiency of skewing, the neural network was proposed to be applied to skewing. A new object detection model with a recurring training strategy called the skewing detection network(SDN), was built by an improved convolutional neural network based on skewing features. Positive sample regression and multi-scale input were added to improve the performance of convolutional networks. To acquire sufficient data on skewing samples for network training, a data augmentation method for skewing was developed, which generated a large number of skewing samples by formularizing the oblique images. The performance of SDN was evaluated by comprehensive comparison experiment. The results show that the SDN performed outstandingly with a precision of 98%,an F-score of 0.97, and the error of skewing slope is controlled at ±8%, better than other target detection models. Compared to YOLOs, the SDN has excellent performance in real skewing sample detection and has good cross-data set detection performance.

参考文献/References:

[1] 史先传, 徐镇冬, 苏胜辉, 等. 整纬设备光电检测头的设计与研究[J]. 传感技术学报, 2019, 32(8): 1169-1174.
SHI Xianchuan, XU Zhendong, SU Shenghui, et al. Design and research on photoelectric detector of weft-straightener[J]. Chinese Journal of Sensors and Actuators, 2019, 32(8): 1169-1174.(in Chinese)
[2] LEE J Y, BAE G H, CHUNG Y S, et al. Analysis and control of camera type weft straightener[J]. World Academy of Science, Engineering and Technology, 2017, 11(3): 567-571.
[3] 王奇锴, 潘如如, 高卫东, 等. 基于图像处理的牛仔织物纬斜检测方法[J]. 棉纺织技术, 2020, 48(6): 31-35.
WANG Qikai, PAN Ruru, GAO Weidong, et al. Test method of denim fabric bias filling based on image processing[J]. Cotton Textile Technology, 2020, 48(6): 31-35.(in Chinese)
[4] 史先传, 董冲, 苏胜辉, 等. 结合Sobel和PPHT的织物纬斜检测方法研究[J]. 计算机测量与控制, 2020, 28(8): 48-52, 57.
SHI Xianchuan, DONG Chong, SU Shenghui, et al. Research on fabric weft skew detection method using Sobel and PPHT[J]. Computer Measurement and Control, 2020, 28(8): 48-52, 57.(in Chinese)
[5] MEI S A, WANG Y D, WEN G J. Automatic fabric defect detection with a multi-scale convolutional denoising autoencoder network model[J]. Sensors, 2018, 18(4): 1064.
[6] ZHAO Z X, LI B, DONG R, et al. A surface defect detection method based on positive samples[C]//Pacific Rim International Conference on Artificial Intelligence. Cham: Springer, 2018: 473- 481.
[7] LIU J H, WANG C Y, SU H, et al. Multistage GAN for fabric defect detection[J]. IEEE Transactions on Image Processing, 2020, 29: 3388-3400.
[8] ZHANG H W, QIAO G H, LU S A, et al. Attention-based Feature Fusion Generative Adversarial Network for yarn-dyed fabric defect detection[J]. Textile Research Journal, 2023, 93(5/6): 1178-1195.
[9] TSAI D M, JEN P H. Autoencoder-based anomaly detection for surface defect inspection[J]. Advanced Enginee-ring Informatics, 2021, 48: 101272.
[10] XU R G, HAO R Y, HUANG B Q. Efficient surface defect detection using self-supervised learning strategy and segmentation network[J]. Advanced Engineering Informatics, 2022, 52: 101566.
[11] FANG J C, LIU Q, LI J Z. A deployment scheme of YOLO v5 with inference optimizations based on the triton inference server[C]//2021 IEEE 6th International Conference on Cloud Computing and Big Data Analytics(ICCCBDA). Chengdu: IEEE, 2021: 441- 445.
[12] 钱炜. 基于神经网络的织物疵点检测研究[D]. 上海: 东华大学, 2018.
[13] 天池. 布匹瑕疵检测数据集[EB/OL].(2020-10-21)[2023-03-30]. https://tianchi.aliyun.com/dataset/dataDetail?dataId=79336.
[14] ZHANG Z. A flexible new technique for camera calibration[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2000, 22(11): 1330-1334.
[15] WERMAN M. Affine invariants[M]//Computer Vision. Cham: Springer International Publishing, 2020: 1-3.
[16] 刘建宝. 面向摄像整纬器的织物纬斜检测与整纬研究[D]. 泉州: 华侨大学, 2020.
[17] SUN C, SHRIVASTAVA A, SINGH S, et al. Revisiting unreasonable effectiveness of data in deep learning era[C]//2017 IEEE International Conference on Computer Vision(ICCV). Venice: IEEE, 2017: 843-852.
[18] DUMOULIN V, VISIN F. A guide to convolution arith-metic for deep learning[EB/OL].(2018-01-12)[2023-03-30]. https://arxiv.org/abs/1603.07285.
[19] VIRAKTAMATH D S, NAVALGI P, NEELOPANT A. Comparison of YOLO v3 and SSD algorithms[J]. International Journal of Engineering Research and Technology, 2021, 10(2): 193-196.
[20] 侯谕融, 狄岚, 梁久祯. 融合高斯金字塔特征的低分辨率人脸识别[J]. 小型微型计算机系统, 2021, 42(10): 2107-2115.
HOU Yurong, DI Lan, LIANG Jiuzhen. Fusion of Gaussian image pyramid feature for low-resolution face recognition[J]. Journal of Chinese Computer Systems, 2021, 42(10): 2107-2115.(in Chinese)
[21] YU J H, JIANG Y N, WANG Z Y, et al. UnitBox: an advanced object detection network[C]//Proceedings of the 24th ACM international conference on Multimedia. New York: ACM, 2016: 516-520.
(责任编辑:沈天琦)

更新日期/Last Update: 2023-10-31