Çelik Yüzeylerdeki Kusurların Tespiti için Topluluk Öğrenme Tabanlı Gelişmiş Segmentasyon Yöntemi

Abstract

Bu çalışmada, çelik yüzeylerdeki kusurların tespiti için topluluk öğrenme tabanlı bir gelişmiş segmentasyon yöntemi sunulmuştur. Günümüzde, çelik üretim süreçlerinde kalite kontrolün önemi artmakta ve kusur tespiti teknolojilerine olan ihtiyaç hızla büyümektedir. Bu bağlamda, çeşitli çelik yüzey kusurlarını doğru bir şekilde tespit edebilen ve bölütleyebilen güçlü bir modele ihtiyaç vardır. Önerilen sistem, derin öğrenme tabanlı çoklu model yaklaşımlarını topluluk öğrenme ile birleştirerek, segmentasyon doğruluğunu önemli ölçüde artırmaktadır. Kullanılan topluluk öğrenme stratejisi, farklı mimarilere sahip birden fazla derin öğrenme modelinin güçlü yönlerini bir araya getirerek, sistemimizin genel performansını iyileştirmekte ve çeşitli kusur türlerine karşı daha duyarlı hale getirmektedir. Bu yaklaşım, geniş bir veri seti üzerinde test edilmiş ve %77,98 mIoU oranı elde ederek mevcut tekil model tabanlı yöntemlere kıyasla önemli bir performans artışı göstermiştir. Sonuçlar, önerilen topluluk öğrenme tabanlı segmentasyon yönteminin, çelik yüzeylerdeki kusurların tespiti konusunda etkili bir çözüm sunduğunu ve endüstriyel kalite kontrol süreçlerinde potansiyel uygulamalara sahip olduğunu ortaya koymaktadır.

Keywords

• Çelik Yüzey Kusur Tespiti
• Topluluk Öğrenme
• Gelişmiş Segmentasyon Yöntemleri

Ensemble Learning Based Advanced Segmentation Method for Detecting Defects on Steel Surfaces

Abstract

In this study, an ensemble learning-based advanced segmentation method is presented for the detection of defects on steel surfaces. Nowadays, the importance of quality control in steel production processes is increasing and the need for defect detection technologies is growing rapidly. In this context, a powerful model that can accurately detect and segment various steel surface defects is needed. The proposed system significantly increases segmentation accuracy by combining deep learning-based multi-model approaches with ensemble learning. The ensemble learning strategy used combines the strengths of multiple deep learning models with different architectures, improving the overall performance of our system and making it more susceptible to various types of defects. This approach has been tested on a large dataset and achieved a mIoU rate of 77.98%, demonstrating a significant performance improvement over existing single model-based methods. The results reveal that the proposed ensemble learning-based segmentation method offers an effective solution for detecting defects on steel surfaces and has potential applications in industrial quality control processes.

Keywords

• Steel Surface Defect Detection
• Ensemble Learning
• Advanced Segmentation Methods.

References

1. [1] Sime, D. M., Wang, G., Zeng, Z., & Peng, B. (2024). Deep learning-based automated steel surface defect segmentation: a comparative experimental study. Multimedia Tools and Applications, 83(1), 2995-3018.
2. [2] Wang, G. Q., Zhang, C. Z., Chen, M. S., Lin, Y. C., Tan, X. H., Kang, Y. X., ... & Zhao, W. W. (2024). A high-accuracy and lightweight detector based on a graph convolution network for strip surface defect detection. Advanced Engineering Informatics, 59, 102280.
3. [3] Tang, B., Chen, L., Sun, W., & Lin, Z. K. (2023). Review of surface defect detection of steel products based on machine vision. IET Image Processing, 17(2), 303-322.
4. [4] Long, Y., Zhang, J., Huang, S., Peng, L., Wang, W., Wang, S., & Zhao, W. (2022). A novel crack quantification method for ultra-high-definition magnetic flux leakage detection in pipeline inspection. IEEE Sensors Journal, 22(16), 16402-16413.
5. [5] Santoro, L., Sesana, R., Molica Nardo, R., & Curá, F. (2023). Infrared in-line monitoring of flaws in steel welded joints: a preliminary approach with SMAW and GMAW processes. The International Journal of Advanced Manufacturing Technology, 128(5), 2655-2670.
6. [6] Akhyar, F., Lin, C. Y., Muchtar, K., Wu, T. Y., & Ng, H. F. (2019, September). High efficient single-stage steel surface defect detection. In 2019 16th IEEE International Conference on Advanced Video and Signal Based Surveillance (AVSS) (pp. 1-4). IEEE.
7. [7] Qu, Y., Wan, B., Wang, C., Ju, H., Yu, J., Kong, Y., & Chen, X. (2023). Optimization algorithm for steel surface defect detection based on PP-YOLOE. Electronics, 12(19), 4161.
8. [8] Akhyar, F., Liu, Y., Hsu, C. Y., Shih, T. K., & Lin, C. Y. (2023). FDD: a deep learning–based steel defect detectors. The International Journal of Advanced Manufacturing Technology, 126(3-4), 1093-1107.

9. [9] Shao, Y., Fan, S., Sun, H., Tan, Z., Cai, Y., Zhang, C., & Zhang, L. (2023). Multi-Scale Lightweight Neural Network for Steel Surface Defect Detection. Coatings, 13(7), 1202.
10. [10] Hussain, M. (2023). YOLO-v1 to YOLO-v8, the Rise of YOLO and Its Complementary Nature toward Digital Manufacturing and Industrial Defect Detection. Machines, 11(7), 677.
11. [11] Üzen, H., Turkoglu, M., Aslan, M., & Hanbay, D. (2023). Depth-wise Squeeze and Excitation Block-based Efficient-Unet model for surface defect detection. The Visual Computer, 39(5), 1745-1764.
12. [12] Yang, L., Xu, S., Fan, J., Li, E., & Liu, Y. (2023). A pixel-level deep segmentation network for automatic defect detection. Expert Systems with Applications, 215, 119388.
13. [13] Zhou, G., & Sun, H. (2020, June). Defect detection method for steel based on semantic segmentation. In 2020 IEEE 5th Information Technology and Mechatronics Engineering Conference (ITOEC) (pp. 975-979). IEEE.
14. [14] Aydin, I., Güçlü, E., & Akin, E. (2023, November). An Improved Unsupervised Convolutional Neural Networks for Detection of Steel Wire Defects. In 2023 International Conference on Innovation and Intelligence for Informatics, Computing, and Technologies (3ICT) (pp. 402-407). IEEE.
15. [15] Xu, R., Hao, R., & Huang, B. (2022). Efficient surface defect detection using self-supervised learning strategy and segmentation network. Advanced Engineering Informatics, 52, 101566..
16. [16] Ronneberger, O., Fischer, P., & Brox, T. (2015). U-net: Convolutional networks for biomedical image segmentation. In Medical image computing and computer-assisted intervention–MICCAI 2015: 18th international conference, Munich, Germany, October 5-9, 2015, proceedings, part III 18 (pp. 234-241). Springer International Publishing..
17. [17] Qin, X., Fan, D. P., Huang, C., Diagne, C., Zhang, Z., Sant'Anna, A. C., ... & Shao, L. (2021). Boundary-aware segmentation network for mobile and web applications. arXiv preprint arXiv:2101.04704.
18. [18] Goncalves, J. P., Pinto, F. A., Queiroz, D. M., Villar, F. M., Barbedo, J. G., & Del Ponte, E. M. (2021). Deep learning architectures for semantic segmentation and automatic estimation of severity of foliar symptoms caused by diseases or pests. Biosystems engineering, 210, 129-142.