Metal Surface Defect Detection Based on IADSA Deep Transfer Network

doi:10.3901/JME.2023.24.046

Abstract

Abstract: Aiming at the problem of low detection accuracy of metal surface defects in the domain shift environment, a metal surface defect detection method based on improved adversarial domain separation and adaptation (IADSA) deep transfer network was proposed. First, the performance evaluation mechanism of IADSA model based on classification loss is established to perceive the training status of the model. The spatial linear interpolation method is proposed to adaptively mine the sample information hidden in the migration space to improve the feature extraction ability of the network. Then, the classification result of new samples is used as the main measurement index to measure the performance of its contribution to the network, and the contribution performance is applied as a weight to the classification loss, which aims to eliminate the influence of noise samples on the model. Finally, dynamic weights are added to optimize adversarial loss and smooth network parameters in the process of adversarial training to improve the discrimination performance of the model. The experimental results show that the proposed method achieves higher detection accuracy of metal surface defects in an unsupervised environment.

Key words: metal surface defect detection, unsupervised domain adaptation, domain adversarial migration, sample mining, dynamic weighting

CLC Number:

TH17

SU Lei, WANG Lijian, QI Yang, ZHANG Siyu, GU Jiefei, LI Ke. Metal Surface Defect Detection Based on IADSA Deep Transfer Network[J]. Journal of Mechanical Engineering, 2023, 59(24): 46-55.

References

[1] 胡慧君,李元香,刘茂福,等. 基于机器学习的带钢表面缺陷分类方法研究[J]. 计算机工程与设计,2014,35(2):620-624. HU Huijun,LI Yuanxiang,LIU Maofu,et al. Steel strip surface defects classification based on machine 1earning[J]. Computer Engineering and Design,2014,35(2):620-624.
[2] FU Guizhong,SUN Peize,ZHU Wenbin,et al. A deep-learning-based approach for fast and robust steel surface defects classification[J]. Optics and Lasers in Engineering,2019,121:397-405.
[3] 孙凯. 基于机器学习的带钢表面缺陷分类[D]. 武汉:华中科技大学,2016. SUN Kai. Machine Learning based surface defect classification for steel strips[D]. Wuhan:Huazhong University of Science and Technology,2016.
[4] ZHAO Jiuliang,PENG Yishu,YAN Yunhui. Steel surface defect classification based on discriminant manifold regularized local descriptor[J]. IEEE Access,2018,6:71719-71731.
[5] HU Huijun,LIU Ya,LIU Maofu,et al. Surface defect classification in large-scale strip steel image collection via hybrid chromosome genetic algorithm[J]. Neurocomputing,2016,181(12):86-95.
[6] GHORAI S,MUKHERJEE A,GANGADARAN M,et al. Automatic defect detection on hot-rolled flat steel products[J]. IEEE Transactions on Instrumentation and Measurement,2013,62(3):612-621.
[7] 蓝金辉,王迪,申小盼. 卷积神经网络在视觉图像检测的研究进展[J]. 仪器仪表学报,2020,41(4):167-182. LAN Jinhui,WANG Di,SHEN Xiaopan. Research progress on visual image detection based on convolutional neural network[J]. Chinese Journal of Scientific Instrument,2020,41(4):167-182.
[8] ZHOU Shiyang,CHEN Youping,ZHANG Dailin,et al. Classification of surface defects on steel sheet using convolutional neural networks[J]. Materials and Technology,2017,51(1):123-131.
[9] CHU Maoxiang,LIU Xiaoping,GONG Rongfen,et al. Multi-class classification method using twin support vector machines with multi-information for steel surface defects[J]. Chemometrics and Intelligent Laboratory Systems,2018,176:108-118.
[10] GONG Rongfen,CHU Maoxiang,YANG Yonghui,et al. A multi-class classifier based on support vector hyperspheres for steel plate surface defects[J]. Chemometrics and Intelligent Laboratory Systems,2019,188(15):70-78.
[11] 余永维,殷国富,殷鹰,等. 基于深度学习网络的射线图像缺陷识别方法[J]. 仪器仪表学报,2014,35(9):2012-2019. YU Yongwei,YIN Guofu,YIN Ying,et al. Defect recognition for radiographic image based on deep learning network[J]. Chinese Journal of Scientific Instrument,2014,35(9):2012-2019.
[12] 李可,祁阳,宿磊,等. 基于改进ACGAN的钢表面缺陷视觉检测方法[J]. 机械工程学报,2022,58(24):32-40. LI Ke,QI Yang,SU Lei,et al. Visual inspection of steel surface defects based on improved auxiliary classification generation adversarial network[J]. Journal of Mechanical Engineering,2022,58(24):32-40.
[13] 姜洪权,贺帅,高建民,等. 一种改进卷积神经网络模型的焊缝缺陷识别方法[J]. 机械工程学报,2020,56(8):235-242. JIANG Hongquan,HE Shuai,GAO Jianming,et al. An improved convolutional neural network for weld defect recognition[J]. Journal of Mechanical Engineering,2020,56(8):235-242.
[14] 周俊. 基于无监督领域适应的图像分类方法研究[D]. 南京:南京邮电大学,2020. ZHOU Jun. Image classification algorithm method based on unsupervised domain adaptation[D]. Nanjing:Nanjing University of Posts and Telecommunications,2020.
[15] ZHANG Siyu,ZHANG Qiuju,GU Jiefei,et al. Visual inspection of steel surface defects based on domain adaptation and adaptive convolutional neural network[J]. Mechanical Systems and Signal Processing,2021,153:107541.
[16] GANIN Y,AJAKAN H,LAROCHELLE H,et al. Domain-adversarial training of neural networks[J]. Journal of Machine Learning Research,2016,17:1-35.
[17] BOUSMALIS K,TRIGEORGIS G,SILBERMAN N,et al. Domain separation networks[C/CD]//Conference on Neural Information Processing Systems,2016.
[18] TSAI J,CHIEN J. Adversarial domain separation and adaptation[C]//2017 IEEE 27th International Workshop on Machine Learning for Signal Processing (MLSP). IEEE,2017:1-6.
[19] DONG Hongwen,SONG Kechen,HE Yu,et al. PGA-Net:Pyramid feature fusion and global context attention network for automated surface defect detection[J]. IEEE Transactions on Industrial Informatics,2020,16(12):7448-7458.
[20] LOSHCHILOV I,HUTTER F. Decoupled weight decay regularization[C]//ICLR 2019,Seventh International Conference on Learning Representations,CMU,MSR,MILA.
[21] MA Guohong,YU Lesheng,YUAN Haitao. A vision-based method for lap weld defects monitoring of galvanized steel sheets using convolutional neural network[J]. Journal of Manufacturing Processes,2021,64:130-139.
[22] ZHOU Qiang,ZHOU Wenan,WANG Shirui,et al. Unsupervised domain adaptation with adversarial distribution adaptation network[J]. Neural Computing and Applications,2020,33(13):7709-7721.
[23] LONG Mingsheng,CAO Yue,CAO Zhangjie,et al. Transferable representation learning with deep adaptation networks[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence,2018,41:3071-3085.