Journal of the Korea Society of Computer and Information (한국컴퓨터정보학회논문지)

Korean Society of Computer Information (한국컴퓨터정보학회)
권호 표지
DOI QR코드

DOI QR Code

YOLOv8-based plastic surface inspector with custom labeling for defect detection

  • In-Bok Jung (Dept. of Biomedical Convergence Engineering, Gangneung-Wonju National University) ;
  • Sangmin Suh (Dept. of Information & Telecommunication Engineering, Gangneung-Wonju National University)
  • Received : 2024.07.31
  • Accepted : 2024.10.29
  • Published : 2024.11.29
Download PDF
⟨ Previous Next ⟩

Abstract

The rapid advancement of society due to industrialization, particularly through mass production enabled by automation, has led to the production of numerous products. However, it is difficult to ensure that all products are manufactured perfectly without defects. Therefore, identifying defects in products during the production process has become crucial. In modern society, detecting defects in various materials is highly valued. This paper focuses on detecting defects in plastic materials, which are among the most widely used and practical materials. In this study, we manually labeled the dataset, creating a dataset consisting of two classes. We utilized the YOLOv8 (You Only Look Once) model, which is capable of object detection, for training. To ensure fair evaluation, k-Fold Cross Validation was performed, resulting in an average F1 Score of 0.95, mAP50 of 0.97, and mAP50-95 of 0.68.

산업화로 인해 사회는 빠르게 발전하고 있다. 특히 자동화에 의한 대량 생산으로 많은 제품이 생산되고 있다. 그러나 모든 제품이 결함 없이 완벽하게 생산되기는 어렵다. 그러므로 생산과정 중에 제품에 생기는 결함을 찾아내는 것은 중요하다. 현대 사회에서는 다양한 소재에서 결함을 찾아내는 것을 중요하게 여기고 있다. 본 논문에서는 가장 활용도가 높고 많이 사용하는 플라스틱에 초점을 맞추어 플라스틱 소재의 결함을 검출하는 것을 목표로 하였다. 본 논문에서는 데이터 세트(Data set)의 레이블링(Labeling)을 직접 하여, 2개의 클래스로 구성된 데이터 세트를 만들었다. 본 논문에서는 객체 탐지가 가능한 YOLOv8(You Only Look Once) 모델을 사용하여 훈련하였다. 공정한 검증을 위해 k-폴드 교차 검증(k-Fold Cross Validation)을 진행하였으며, 평균 F1 Score=0.95, mAP50=0.97, 그리고 mAP50-95=0.68을 얻었다.

Keywords

Acknowledgement

This work was supported by Innovative Human Resource Development for Local Intellectualization program through the Institute of Information & Communications Technology Planning & Evaluation(IITP) grant funded by the Korea government(MSIT)(IITP-2024-RS-2023-00260267)

References

  1. M. I. B. Roslan, Z. Ibrahim, and Z. A. Aziz, "Real-Time Plastic Surface Defect Detection Using Deep Learning," in 2022 IEEE 12th Symposium on Computer Applications & Industrial Electronics (ISCAIE), 2022.
  2. Alaa Aldein M. S. Ibrahim and Jules-Raymond Tapamo, "A Survey of Vision-Based Methods for Surface Defects' Detection and Classification in Steel Products," Informatics, Vol. 11, No. 2, p. 25, 2024.
  3. Z. Ren, F. Fang, N. Yan, and Y. Wu, "State of the Art in Defect Detection Based on Machine Vision," International Journal of Precision Engineering and Manufacturing-Green Technology, Vol. 9, pp. 661-691, 2022. https://doi.org/10.1007/s40684-021-00343-6
  4. R. V. Patil, Y. P. Reddy, and A. M. Thote, "Multi-class Weld Defect Detection and Classification by Support Vector Machine and Artificial Neural Network," Journal of Intelligent Manufacturing, Vol. 31, No. 3, pp. 540-550, Mar. 2020.
  5. F. Nagata, K. Tokuno, K. Mitarai, A. Otsuka, T. Ikeda, and H. Ochi, "Defect detection method using deep convolutional neural network, support vector machine and template matching techniques," Artificial Life and Robotics, Vol. 23, No. 1, pp. 120-130, Jan. 2018.
  6. Y. Zhang, X. Chen, H. Li, J. Zhong, and S. Chen, "Surface defect detection of industrial components based on vision," Journal of Intelligent Manufacturing, Vol. 33, No. 4, pp. 867-879, Apr. 2022. https://doi.org/10.1007/s10845-020-01686-8
  7. L. A. Demidova, "Two-Stage Hybrid Data Classifiers Based on SVM and kNN Algorithms," Symmetry, Vol. 13, No. 4, pp. 615, Apr. 2021.
  8. H. Khanzadeh, J. Hu, S. Azizi, and S. Hashemzadeh, "Machine learning algorithms for defect detection in metal laser-based manufacturing," ScienceDirect, Vol. 28, No. 3, pp. 459-472, Mar. 2020.
  9. M. Kazmi, B. Hafeez, H. R. Khan, and S. A. Qazi, "Machine-Vision-Based Plastic Bottle Inspection for Quality Assurance," in Proceedings of the Engineering Proceedings, Vol. 20, p. 9, MDPI, July 2022.
  10. J. Liang, R. Kong, R. Ma, J. Zhang, and X. Bian, "Aluminum Surface Defect Detection Algorithm Based on Improved YOLOv5," Adv. Theory Simul., Vol. 7, No. 2, pp. 1-11, Dec. 2023, DOI: 10.1002/adts.202300695.
  11. Y. Cao, D. Pang, Q. Zhao, Y. Yan, Y. Jiang, C. Tian, F. Wang, and J. Li, "Improved YOLOv8-GD deep learning model for defect detection in electroluminescence images of solar photovoltaic modules," Engineering Applications of Artificial Intelligence, Vol. 131, p. 107866, 2024, DOI: 10.1016/j.engappai.2024.107866.
  12. M. T. Celik, S. Arslankaya, and A. Yildiz, "Real-time detection of plastic part surface defects using deep learning-based object detection model," Measurement, Vol. 235, No. 114975, pp. 1-14, May 2024, DOI: 10.1016/j.measurement.2024.114975.
  13. C. Yao, L. Wang, Q. Wang, Z. Liu, G. Liu, and M. Zhang, "Deep-Learning-Guided Evaluation Method for the High-Volume Preparation of Flexible Sensors Based on Inkjet Printing," ACS Applied Materials & Interfaces, Vol. 16, No. 10, pp. 13326-13334, Mar. 2024, DOI: 10.1021/acsami.4c00322.
  14. J. U. Ko, J. Lee, T. Kim, Y. C. Kim, and B. D. Youn, "Deep-learning-based fault detection and recipe optimization for a plastic injection molding process under the class-imbalance problem," J. Comput. Des. Eng., Vol. 10, pp. 694-710, Feb. 2023, DOI: 10.1093/jcde/qwad011.
  15. H. Lv, H. Zhang, M. Wang, J. Xu, X. Li, and C. Liu, "Hyperspectral Imaging Based Nonwoven Fabric Defect Detection Method Using LL-YOLOv5," IEEE Access, Vol. 12, pp. 41988-41998, Mar. 2024, DOI: 10.1109/ACCESS.2024.3378739.
  16. G. Jocher, A. Stoken, J. Chaurasia, and A. Borovec, "ultralytics/ultralytics: YOLOv8,", https://github.com/ultralytics/ultralytics.
  17. Nam Trng, Seg_Cass_Crack(Crack) Computer Vision Project, https://universe.roboflow.com/nam-trng/seg_cass_crack-crack.