DOI QR코드

DOI QR Code

Comparison and Application of Deep Learning-Based Anomaly Detection Algorithms for Transparent Lens Defects

딥러닝 기반의 투명 렌즈 이상 탐지 알고리즘 성능 비교 및 적용

  • Received : 2023.09.27
  • Accepted : 2023.12.26
  • Published : 2024.03.31

Abstract

Deep learning-based computer vision anomaly detection algorithms are widely utilized in various fields. Especially in the manufacturing industry, the difficulty in collecting abnormal data compared to normal data, and the challenge of defining all potential abnormalities in advance, have led to an increasing demand for unsupervised learning methods that rely on normal data. In this study, we conducted a comparative analysis of deep learning-based unsupervised learning algorithms that define and detect abnormalities that can occur when transparent contact lenses are immersed in liquid solution. We validated and applied the unsupervised learning algorithms used in this study to the existing anomaly detection benchmark dataset, MvTecAD. The existing anomaly detection benchmark dataset primarily consists of solid objects, whereas in our study, we compared unsupervised learning-based algorithms in experiments judging the shape and presence of lenses submerged in liquid. Among the algorithms analyzed, EfficientAD showed an AUROC and F1-score of 0.97 in image-level tests. However, the F1-score decreased to 0.18 in pixel-level tests, making it challenging to determine the locations where abnormalities occurred. Despite this, EfficientAD demonstrated excellent performance in image-level tests classifying normal and abnormal instances, suggesting that with the collection and training of large-scale data in real industrial settings, it is expected to exhibit even better performance.

Keywords

References

  1. Akcay, S., Ameln, D., Vaidya, A., Lakshmanan, B., Ahuja, N., and Genc, U., Anomalib: A deep learning library for anomaly detection, International Conference on Image Processing, 2022, pp. 1706-1710.
  2. Batzner, Kilian, Lars Heckler, and Rebecca Konig, Efficientad: Accurate visual anomaly detection at millisecond-level latencies, ArXiv: 2303.14535, 2023.
  3. Bergmann, P., Fauser, M., Sattlegger, D., and Steger, C., MVTec AD-A comprehensive real-world dataset for unsupervised anomaly detection, Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2019, pp. 9592-9600.
  4. COHEN, Niv; HOSHEN, and Yedid, Sub-image anomaly detection with deep pyramid correspondences, ArXiv: 2005.02357, 2020.
  5. Cui, Y., Liu, Z., and Lian, S., A Survey on Unsupervised Anomaly Detection Algorithms for Industrial Images, IEEE Access, 2023.
  6. Defard, T., Setkov, A., Loesch, A., and Audigier, R., Padim: A Patch Distribution Modeling Framework for Anomaly Detection and Localization, International Conference on Pattern Recognition, 2021, pp. 475-489.
  7. Deng, H. and Li, X., Anomaly detection via reverse distillation from one-class embedding, Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2021, pp. 9737-9746.
  8. Ehret, T., Davy, A., Morel, J. M., and Delbracio, M., Image anomalies: A review and synthesis of detection methods, Journal of Mathematical Imaging and Vision, 2019, Vol. 61, pp. 710-743. https://doi.org/10.1007/s10851-019-00885-0
  9. Gudovskiy, D., Ishizaka, S., and Kozuka, K., Cflow-ad: Real-time unsupervised anomaly detection with localization via conditional normalizing flows, Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision, 2022, pp. 98-107.
  10. Guo, Y., Zeng, Y., Gao, F., Qiu, Y., Zhou, X., Zhong, L., and Zhan, C., Improved YOLOv4-CSP algorithm for detection of bamboo surface sliver defects with extreme aspect ratio, IEEE Access, 2022, Vol. 10, pp. 29810-29820. https://doi.org/10.1109/ACCESS.2022.3152552
  11. Hao, R., Lu, B., Cheng, Y., Li, X., and Huang, B., A Steel Surface Defect Inspection Approach Towards Smart Industrial Monitoring, Journal of Intelligent Manufacturing, 2021, Vol. 32, pp. 1833-1843. https://doi.org/10.1007/s10845-020-01670-2
  12. https://www.mvtec.com/company/research/datasets/mvtec-ad (2022.11.26 access).
  13. Kim, G.N., Kim, S.H., Joo, I., and Yoo, K.H., Detection of Color Contact Lens Defects using Various CNN Models, Journal of The Korea Contents Association, 2022, Vol. 22, No. 12, pp. 160-170. https://doi.org/10.5392/JKCA.2022.22.12.160
  14. Kim, Y.D., Kim, N.K., and Wang, G.N., Determination of Defective Products based on DBSCAN using Temperature Data of Manufacturing Sites, The Korean Society of Manufacturing Technology Engineers, 2020, pp. 126-126.
  15. Kingma, D.P. and Welling, M., Auto-encoding variational bayes, ArXiv:1312.6114, 2013.
  16. Lugaresi, C., Tang, J., Nash, H., McClanahan, C., Uboweja, E., Hays, M., and Grundmann, M., Mediapipe: A framework for building perception pipelines, ArXiv:1906.08172, 2019.
  17. Marco Rudolph, Tom Wehrbein, Bodo Rosenhahn, and Bastian Wandt. Asymmetric student-teacher networks for industrial anomaly detection, Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision, 2023, pp. 2592-2602.
  18. Mishra, P., Verk, R., Fornasier, D., Piciarelli, C., and Foresti, G.L., VT-ADL: A vision transformer network for image anomaly detection and localization, IEEE 30th International Symposium on Industrial Electronics (ISIE), 2021, pp. 01-06.
  19. Pang, G., Shen, C., Cao, L., and Hengel, A.V.D., Deep learning for anomaly detection: A review, ACM Computing Surveys (CSUR), 2021, Vol. 54, No. 2, pp. 1-38. https://doi.org/10.1145/3439950
  20. Redmon, J., Divvala, S., Girshick, R., and Farhadi, A., You only look once: Unified, real-time object detection, Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2016, pp. 779-788.
  21. Roth, K., Pemula, L., Zepeda, J., Scholkopf, B., Brox, T., and Gehler, P., Towards total recall in industrial anomaly detection, Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2022, pp. 14318-14328.
  22. Rudolph, M., Wandt, B., and Rosenhahn, B., Same same but differnet: Semi-supervised defect detection with normalizing flows, Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision, 2021, pp. 1907-1916.
  23. Son, J.H. and Kim, C.O., A Study on the Application of Deep Learning Models for Real-time Defect Detection in the Manufacturing Process - Cases of Defect detection in the Label Printing Process, Journal of Korea Technical Association of the Pulp and Paper Industry, 2021, Vol. 53, No. 5, pp. 74-81. https://doi.org/10.7584/JKTAPPI.2021.10.53.5.74
  24. Um, I.S., Jeong, J.H., and Choi, Y.J., Root Cause Analysis and Process Condition Optimization of MEA Manufacturing Systems Using XAI and Bayesian Optimization, Korean Institute of Industrial Engineers, 2023, pp. 2205-2209.
  25. Wang, G., Han, S., Ding, E., and Huang, D., Student-teacher feature pyramid matching for anomaly detection, ArXiv:2013.04257, 2021.
  26. Yang, J., Li, S., Wang, Z., Dong, H., Wang, J., and Tang, S., Using Deep Learning to Detect Defects in Manufacturing: A Comprehensive Survey and Current Challenges, Materials, 2020, Vol. 13, No. 24, p. 5755.
  27. Yu, J., Zheng, Y., Wang, X., Li, W., Wu, Y., Zhao, R., and Wu, L., Fastflow: Unsupervised anomaly detection and localization via 2d normalizing flows, ArXiv:2111.07677, 2021.
  28. Zavrtanik, V., Kristan, M., and Skocaj, D., Draem-a discriminatively trained reconstruction embedding for surface anomaly detection, Proceedings of the IEEE/CVF International Conference on Computer Vision, 2021, pp. 8330-8339.