Mostrar el registro sencillo del documento

dc.rights.licenseAtribución-NoComercial-SinDerivadas 4.0 Internacional
dc.contributor.advisorPrieto Ortiz, Flavio Augusto
dc.contributor.authorKicker, Claudia
dc.date.accessioned2022-08-26T20:21:33Z
dc.date.available2022-08-26T20:21:33Z
dc.date.issued2022
dc.identifier.urihttps://repositorio.unal.edu.co/handle/unal/82144
dc.descriptionilustraciones, graficas
dc.description.abstractAnomaly detection is of great importance in the production of steel plates, in order to guarantee that the products are defect-free. Various deep-learning approaches for defect-detection in steel surfaces have emerged in the recent years, however, they are mainly limited to plain steel surfaces. Furthermore, deep-learning-based anomaly detection is still a challenging task if not enough training samples are available, which is often the case in real world scenarios. As for patterned steel plates, the availability anomalous samples is low, as productions are optimized to minimize the occurrence of defects. Hence, the main purpose of this work is the determination of a suitable deep learning-based method for the detection of surface anomalies in patterned steel plates. Several methods were trained and compared in terms of segmentation ability and classification accuracy. On the one hand, a convolutional neural network pretrained on artificial defects was adapted to images from a different production line, of which only anomaly-free data was available for training. On the other hand, an autoencoder was trained in a semi-supervised fashion to reconstruct anomaly-free images, in order to identify defective regions by measuring the reconstruction error. Moreover, an analysis of the frequency spectrum for images of patterned steel plates under the application of discrete fourier transform is provided. It was found out that a reconstructing autoencoder trained with a structural similarity loss provided the most accurate localizations of surface anomalies.
dc.description.abstractLa detección de anomalías es de gran importancia en la producción de placas de acero para garantizar que los productos no tengan defectos. En los últimos años han surgido varios métodos de aprendizaje profundo para la detección de defectos en superficies de acero limitándose principalmente a superficies de acero planas. Además, la detección de anomalías basada en el aprendizaje profundo sigue siendo una tarea difícil si no se dispone de suficientes muestras de entrenamiento, lo que suele ocurrir en escenarios del mundo real. En cuanto a las placas de acero texturizadas, como las láminas alfajor, la disponibilidad de muestras anómalas es baja, ya que las producciones están optimizadas para minimizar la aparición de defectos. Por lo tanto, el objetivo principal de este trabajo es la determinación de un método adecuado basado en el aprendizaje profundo, para la detección de anomalías superficiales en placas de acero texturizadas. Se entrenaron varios modelos, los que se compararon en términos de capacidad de segmentación y precisión de clasificación. Por un lado, se adaptó una red neuronal convolucional pre-entrenada en defectos artificiales a imágenes procedentes de una línea de producción diferente, de la que solo se disponía de datos libres de anomalías para su entrenamiento. Por otro lado, se entrenó un autocodificador de forma semi-supervisada para reconstruir imágenes libres de anomalías, con el fin de identificar las regiones defectuosas midiendo el error de reconstrucción. Además, se realiza un análisis del espectro de frecuencias para las imágenes de placas de acero texturizadas bajo la aplicación de la transformada discreta de Fourier. Se descubrió que un autocodificador de reconstrucción entrenado con una función de pérdida que mide la similitud estructural, proporciona las localizaciones más precisas de las anomalías superficiales. (Texto tomado de la fuente)
dc.format.extentxvi, 48 páginas
dc.format.mimetypeapplication/pdf
dc.language.isoeng
dc.publisherUniversidad Nacional de Colombia
dc.rights.urihttp://creativecommons.org/licenses/by-nc-nd/4.0/
dc.subject.ddc670 - Manufactura::672 - Hierro, acero, otras aleaciones ferrosas
dc.titleAutomated defect detection approach for production processes of patterned steel plates using computer vision and deep learning
dc.typeTrabajo de grado - Maestría
dc.type.driverinfo:eu-repo/semantics/masterThesis
dc.type.versioninfo:eu-repo/semantics/acceptedVersion
dc.publisher.programBogotá - Ingeniería - Maestría en Ingeniería - Ingeniería Mecánica
dc.description.degreelevelMaestría
dc.description.degreenameMaestría en Ingeniería - Ingeniería Mecánica
dc.description.researchareaAutomation, Control and Mechatronics
dc.identifier.instnameUniversidad Nacional de Colombia
dc.identifier.reponameRepositorio Institucional Universidad Nacional de Colombia
dc.identifier.repourlhttps://repositorio.unal.edu.co/
dc.publisher.departmentDepartamento de Ingeniería Mecánica y Mecatrónica
dc.publisher.facultyFacultad de Ingeniería
dc.publisher.placeBogotá, Colombia
dc.publisher.branchUniversidad Nacional de Colombia - Sede Bogotá
dc.relation.indexedRedCol
dc.relation.indexedLaReferencia
dc.relation.referencesAtienza, Rowel: Advanced Deep Learning with Keras: Apply deep learning techniques, autoencoders, GANs, variational autoencoders, deep reinforcement learning, policy gradients, and more. 1. Birmingham : Packt Publishing Limited, 2018. – ISBN 9781788624534
dc.relation.referencesBergmann, Paul ; Batzner, Kilian ; Fauser, Michael ; Sattlegger, David ; Steger, Carsten: The MVTec Anomaly Detection Dataset: A Comprehensive Real-World Dataset for Unsupervised Anomaly Detection. In: International Journal of Computer Vision 129 (2021), No. 4, pp. 1038–1059. – ISSN 0920–5691
dc.relation.referencesBergmann, Paul ; Fauser, Michael ; Sattlegger, David ; Steger, Carsten: MVTec AD — A Comprehensive Real-World Dataset for Unsupervised Anomaly Detection. In: 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), IEEE, 2019. – ISBN 978–1–7281–3293–8, pp. 9584–9592
dc.relation.referencesBergmann, Paul ; Fauser, Michael ; Sattlegger, David ; Steger, Carsten: Uninformed Students: Student-Teacher Anomaly Detection With Discriminative Latent Embeddings. In: 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), IEEE, 2020. – ISBN 978–1–7281–7168–5, pp. 4182–4191
dc.relation.referencesBergmann, Paul ; Löwe, Sindy ; Fauser, Michael ; Sattlegger, David ; Steger, Carsten: Improving Unsupervised Defect Segmentation by Applying Structural Similarity to Autoencoders. (2019), pp. 372–380
dc.relation.referencesBožič, Jakob ; Tabernik, Domen ; Skočaj , Danijel: End-to-end training of a twostage neural network for defect detection. In: 2020 25th International Conference on Pattern Recognition (ICPR), 2021, pp. 5619–5626
dc.relation.referencesBreiman, Leo: Random Forests. In: Machine Learning 45 (2001), No. 1, pp. 5–32. – ISSN 08856125
dc.relation.referencesChandola, Varun ; Banerjee, Arindam ; Kumar, Vipin: Anomaly detection. In: ACM Computing Surveys 41 (2009), No. 3, pp. 1–58. – ISSN 0360–0300
dc.relation.referencesSzegedy, Christian ; Ioffe, Sergey ; Vanhoucke, Vincent ; Alemi, Alexander A. : Inception-v4, Inception-ResNet and the Impact of Residual Connections on Learning. In: Proceedings of the Thirty-First AAAI Conference on Artificial Intelligence (AAAI-17) (2017), pp. 4279–4284
dc.relation.referencesDawson-Howe, Kenneth: A practical introduction to computer vision with OpenCV. Online-edition. Chichester, England : Wiley, 2014. – ISBN 9781118848784
dc.relation.referencesDeutsches Institut für Normung e.V. : DIN EN 10363:2016 D: Continuously hot-rolled patterned steel strip and plate/sheet cut from wide strip - Tolerances on dimensions and shape. October 2016
dc.relation.referencesDi, He ; Ke, Xu ; Peng, Zhou ; Dongdong, Zhou: Surface defect classification of steels with a new semi-supervised learning method. In: Optics and Lasers in Engineering 117 (2019), pp. 40–48. – ISSN 01438166
dc.relation.referencesDi He ; Xu, Ke ; Wang, Dadong: Design of multi-scale receptive field convolutional neural network for surface inspection of hot rolled steels. In: Image and Vision Computing 89 (2019), pp. 12–20. – ISSN 02628856
dc.relation.referencesGlassner, Andrew: Deep Learning: A Visual Approach. No Starch Press, 2021. – ISBN 9781718500730
dc.relation.referencesGonzalez, Rafael C. ; Woods, Richard E.: Digital image processing. Fourth, global edition. New York, New York : Pearson Education, 2018. – ISBN 9781292223070
dc.relation.referencesHan, Lin ; Chen, Yuanhao ; Li, Jiaming ; Zhong, Bowei ; Lei, Yuzhu ; Sun, Minghui: Liver segmentation with 2.5D perpendicular UNets. In: Computers & Electrical Engineering 91 (2021), p. 107118. – ISSN 00457906
dc.relation.referencesHanbay, Kazım ; Talu, Muhammed F. ; Özgüven , Ömer F.: Fabric defect detection systems and methods—A systematic literature review. In: Optik 127 (2016), No. 24, pp. 11960–11973. – ISSN 00304026
dc.relation.referencesHe, Yu ; Song, Kechen ; Meng, Qinggang ; Yan, Yunhui: An End-to-End Steel Surface Defect Detection Approach via Fusing Multiple Hierarchical Features. In: IEEE Transactions on Instrumentation and Measurement 69 (2020), No. 4, pp. 1493–1504. – ISSN 0018–9456
dc.relation.referencesHuang, Yibin ; Qiu, Congying ; Wang, Xiaonan ; Wang, Shijun ; Yuan, Kui: A Compact Convolutional Neural Network for Surface Defect Inspection. In: Sensors (Basel, Switzerland) 20 (2020), No. 7
dc.relation.referencesHuang, Zheng ; Wu, Jiajun ; Xie, Feng: Automatic surface defect segmentation for hot-rolled steel strip using depth-wise separable U-shape network. In: Materials Letters 301 (2021), p. 130271. – ISSN 0167577X
dc.relation.referencesLappas, Demetris ; Argyriou, Vasileios ; Makris, Dimitrios: Fourier Transformation Autoencoders for Anomaly Detection. In: ICASSP 2021 - 2021 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), IEEE, 2021. – ISBN 978–1–7281–7605–5, pp. 1475–1479
dc.relation.referencesLi, Yundong ; Zhao, Weigang ; Pan, Jiahao: Deformable Patterned Fabric Defect Detection With Fisher Criterion-Based Deep Learning. In: IEEE Transactions on Automation Science and Engineering 14 (2017), No. 2, pp. 1256–1264. – ISSN 1545– 5955
dc.relation.referencesLi, Yuyuan ; Zhang, Dong ; Lee, Dah-Jye: Automatic fabric defect detection with a wide-and-compact network. In: Neurocomputing 329 (2019), pp. 329–338. – ISSN 09252312
dc.relation.referencesLin, Hui ; Li, Bin ; Wang, Xinggang ; Shu, Yufeng ; Niu, Shuanglong: Automated defect inspection of LED chip using deep convolutional neural network. In: Journal of Intelligent Manufacturing 30 (2019), No. 6, pp. 2525–2534. – ISSN 0956–5515
dc.relation.referencesLiu, Jie ; Song, Kechen ; Feng, Mingzheng ; Yan, Yunhui ; Tu, Zhibiao ; Zhu, Liu: Semi-supervised anomaly detection with dual prototypes autoencoder for industrial surface inspection. In: Optics and Lasers in Engineering 136 (2021), p. 106324. – ISSN 01438166
dc.relation.referencesLiu, Yang ; Xu, Ke ; Xu, Jinwu: Periodic Surface Defect Detection in Steel Plates Based on Deep Learning. In: Applied Sciences 9 (2019), No. 15, p. 3127
dc.relation.referencesLuo, Qiwu ; Liu, Kexin ; Su, Jiaojiao ; Yang, Chunhua ; Gui, Weihua ; Liu, Li ; Silven, Olli: Waterdrop Removal From Hot-Rolled Steel Strip Surfaces Based on Progressive Recurrent Generative Adversarial Networks. In: IEEE Transactions on Instrumentation and Measurement 70 (2021), pp. 1–11. – ISSN 0018–9456
dc.relation.referencesMei, Shuang ; Wang, Yudan ; Wen, Guojun: Automatic Fabric Defect Detection with a Multi-Scale Convolutional Denoising Autoencoder Network Model. In: Sensors (Basel, Switzerland) 18 (2018), No. 4
dc.relation.referencesMueller, John P. ; Massaron, Luca: Deep learning for dummies. Hoboken, NJ : John Wiley & Sons Inc, 2019 (For dummies). – ISBN 9781119543039
dc.relation.referencesNapoletano, Paolo ; Piccoli, Flavio ; Schettini, Raimondo: Anomaly Detection in Nanofibrous Materials by CNN-Based Self-Similarity. In: Sensors (Basel, Switzerland) 18 (2018), No. 1
dc.relation.referencesNapoletano, Paolo ; Piccoli, Flavio ; Schettini, Raimondo: Semi-supervised anomaly detection for visual quality inspection. In: Expert Systems with Applications 183 (2021), p. 115275. – ISSN 09574174
dc.relation.referencesNeogi, Nirbhar ; Mohanta, Dusmanta K. ; Dutta, Pranab K.: Review of vision-based steel surface inspection systems. In: EURASIP Journal on Image and Video Processing 2014 (2014), No. 1
dc.relation.referencesNgan, Henry Y. ; Pang, Grantham K. ; Yung, Nelson H.: Automated fabric defect detection—A review. In: Image and Vision Computing 29 (2011), No. 7, pp. 442–458. – ISSN 02628856
dc.relation.referencesPaulraj, M. P. ; Shukry, A. M. M. ; Yaacob, S. ; Adom, A. H. ; Krishnan, R. P.: Structural steel plate damage detection using DFT spectral energy and artificial neural network. In: 2010 6th International Colloquium on Signal Processing & its Applications, IEEE, 2010. – ISBN 978–1–4244–7121–8, pp. 1–6
dc.relation.referencesPlanche, Benjamin: Hands-On Computer Vision with TensorFlow 2: Leverage deep learning to create powerful image processing apps with TensorFlow 2.0 and Keras. 1. Birmingham : Packt Publishing Limited, 2019. – ISBN 9781788839266
dc.relation.referencesRacki, Domen ; Tomazevic, Dejan ; Skocaj, Danijel: A Compact Convolutional Neural Network for Textured Surface Anomaly Detection. In: 2018 IEEE Winter Conference on Applications of Computer Vision (WACV), IEEE, 32018. – ISBN 978– 1–5386–4886–5, pp. 1331–1339
dc.relation.referencesRedmon, Joseph ; Divvala, Santosh ; Girshick, Ross ; Farhadi, Ali: You Only Look Once: Unified, Real-Time Object Detection. In: 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), IEEE, 2016
dc.relation.referencesRen, Ruoxu ; Hung, Terence ; Tan, Kay C.: A Generic Deep-Learning-Based Approach for Automated Surface Inspection. In: IEEE transactions on cybernetics 48 (2018), No. 3, pp. 929–940
dc.relation.referencesRonneberger, Olaf ; Fischer, Philipp ; Brox, Thomas: U-Net: Convolutional Networks for Biomedical Image Segmentation. In: Navab, Nassir (Ed. ) ; Hornegger, Joachim (Ed. ) ; Wells, William M. (Ed. ) ; Frangi, Alejandro F. (Ed. ): Medical Image Computing and Computer-Assisted Intervention – MICCAI 2015. Cham : Springer International Publishing, 2015. – ISBN 978–3–319–24574–4, pp. 234–241
dc.relation.referencesSchlegl, Thomas ; Seeböck , Philipp ; Waldstein, Sebastian M. ; Langs, Georg ; Schmidt-Erfurth, Ursula: f-AnoGAN: Fast unsupervised anomaly detection with generative adversarial networks. In: Medical image analysis 54 (2019), pp. 30–44
dc.relation.referencesSchlegl, Thomas ; Seeböck, Philipp ; Waldstein, Sebastian M. ; Schmidt-Erfurth, Ursula ; Langs, Georg: Unsupervised Anomaly Detection with Generative Adversarial Networks to Guide Marker Discovery, Springer, Cham, 2017, pp. 146–157
dc.relation.referencesSee, Judi E. ; Drury, Colin G. ; Speed, Ann ; Williams, Allison ; Khalandi, Negar: The Role of Visual Inspection in the 21 st Century. In: Proceedings of the Human Factors and Ergonomics Society Annual Meeting 61 (2017), No. 1, pp. 262–266. – ISSN 2169–5067
dc.relation.referencesSong, Kechen ; Yan, Yunhui: A noise robust method based on completed local binary patterns for hot-rolled steel strip surface defects. In: Applied Surface Science 285 (2013), pp. 858–864. – ISSN 01694332
dc.relation.referencesSoulami, Khaoula B. ; Kaabouch, Naima ; Saidi, Mohamed N. ; Tamtaoui, Ahmed: Breast cancer: One-stage automated detection, segmentation, and classification of digital mammograms using UNet model based-semantic segmentation. In: Biomedical Signal Processing and Control 66 (2021), p. 102481. – ISSN 17468094
dc.relation.referencesSteinwart, Ingo ; Christmann, Andreas: Support vector machines. New York, NY : Springer, 2008 (Information science and statistics). – ISBN 978–0–387–77241–7
dc.relation.referencesTabernik, Domen ; Šela, Samo ; Skvarč, Jure ; Skočaj , Danijel: Segmentation- based deep-learning approach for surface-defect detection. In: Journal of Intelligent Manufacturing 31 (2020), No. 3, pp. 759–776. – ISSN 0956–5515
dc.relation.referencesTsai, Du-Ming ; Jen, Po-Hao: Autoencoder-based anomaly detection for surface defect inspection. In: Advanced Engineering Informatics 48 (2021), p. 101272. – ISSN 14740346
dc.relation.referencesWang, Lu ; Zhang, Dongkai ; Guo, Jiahao ; Han, Yuexing: Image Anomaly Detection Using Normal Data Only by Latent Space Resampling. In: Applied Sciences 10 (2020), No. 23, p. 8660
dc.relation.referencesWang, Tian ; Chen, Yang ; Qiao, Meina ; Snoussi, Hichem: A fast and robust convolutional neural network-based defect detection model in product quality control. In: The International Journal of Advanced Manufacturing Technology 94 (2018), No. 9-12, pp. 3465–3471. – ISSN 0268–3768
dc.relation.referencesWang, Z. ; Simoncelli, E. P. ; Bovik, A. C.: Multiscale structural similarity for image quality assessment. In: The Thrity-Seventh Asilomar Conference on Signals, Systems & Computers, 2003, IEEE, 2003. – ISBN 0–7803–8104–1, pp. 1398–1402
dc.relation.referencesWei, Bing ; Hao, Kuangrong ; Tang, Xue-song ; Ding, Yongsheng: A new method using the convolutional neural network with compressive sensing for fabric defect classification based on small sample sizes. In: Textile Research Journal 89 (2019), No. 17, pp. 3539–3555. – ISSN 0040–5175
dc.relation.referencesWeimer, Daniel ; Scholz-Reiter, Bernd ; Shpitalni, Moshe: Design of deep convolutional neural network architectures for automated feature extraction in industrial inspection. In: CIRP Annals 65 (2016), No. 1, pp. 417–420. – ISSN 00078506
dc.relation.referencesXie, Xianghua: A Review of Recent Advances in Surface Defect Detection using Texture analysis Techniques. In: ELCVIA Electronic Letters on Computer Vision and Image Analysis 7 (2008), No. 3, p. 1
dc.relation.referencesZhang, Defu ; Song, Kechen ; Xu, Jing ; He, Yu ; Yan, Yunhui: Unified detection method of aluminium profile surface defects: Common and rare defect categories. In: Optics and Lasers in Engineering 126 (2020), p. 105936. – ISSN 01438166
dc.relation.referencesZhang, Hong-wei ; Zhang, Ling-jie ; Li, Peng-fei ; Gu: Yarn-dyed Fabric Defect Detection with YOLOV2 Based on Deep Convolution Neural Networks. In: 2018 IEEE 7th Data Driven Control and Learning Systems Conference (DDCLS), IEEE, 2018
dc.relation.referencesZhang, Yifei: A better autoencoder for image: Convolutional autoencoder, Australian National University, Dissertation
dc.relation.referencesZhou, Shiyang ; Chen, Youping ; Zhang, Dailin ; Xie, Jingming ; Zhou, Yunfei: Classification of surface defects on steel sheet using convolutional neural networks. In: Materiali in tehnologije 51 (2017), No. 1, pp. 123–131. – ISSN 15802949
dc.relation.referencesZimek, Arthur ; Schubert, Erich: Outlier Detection. In: Liu, Ling (Ed. ) ; Özsu, M. T. (Ed. ): Encyclopedia of Database Systems. New York, NY : Springer New York, 2017. – ISBN 978–1–4899–7993–3, pp. 1–5
dc.rights.accessrightsinfo:eu-repo/semantics/openAccess
dc.subject.lembLAMINAS DE HIERRO Y ACERO
dc.subject.lembPlates, iron and steel
dc.subject.proposalDeep learning
dc.subject.proposalAnomaly detection
dc.subject.proposalAutoencoders
dc.subject.proposalCNN
dc.subject.proposalStructural similarity
dc.subject.proposalAprendizaje profundo
dc.subject.proposalDetección de anomalías
dc.subject.proposalAutocodificador
dc.subject.proposalRed neuronal convolucional
dc.subject.proposalSimilitud estructural
dc.title.translatedMétodo para la detección automatizada de defectos en la producción de láminas de acero alfajor mediante visión artificial y aprendizaje profundo
dc.type.coarhttp://purl.org/coar/resource_type/c_bdcc
dc.type.coarversionhttp://purl.org/coar/version/c_ab4af688f83e57aa
dc.type.contentText
dc.type.redcolhttp://purl.org/redcol/resource_type/TM
oaire.accessrightshttp://purl.org/coar/access_right/c_abf2
dcterms.audience.professionaldevelopmentEstudiantes
dcterms.audience.professionaldevelopmentInvestigadores
dcterms.audience.professionaldevelopmentMaestros


Archivos en el documento

Thumbnail

Este documento aparece en la(s) siguiente(s) colección(ones)

Mostrar el registro sencillo del documento

Atribución-NoComercial-SinDerivadas 4.0 InternacionalEsta obra está bajo licencia internacional Creative Commons Reconocimiento-NoComercial 4.0.Este documento ha sido depositado por parte de el(los) autor(es) bajo la siguiente constancia de depósito