Academic Journal of Computing & Information Science, 2023, 6(3); doi: 10.25236/AJCIS.2023.060302.
Heng Xiao1,2, Changwei Zhao1,2, Zhiyong Zhang1,2
1College of Information Engineering, Henan University of Science and Technology, Luoyang, China
2Henan International Joint Laboratory of Cyberspace Security Applications, Henan University of Science and Technology, Luoyang, China
Metal products are indispensable raw materials for society nowadays. Surface defects is inevitably occur during the production process, so defect detection is an important method to improve product quality. For the problem of lack of annotation of defect images in new production lines and low efficiency of manual annotation, we propose a semi-supervised defect detection method with improvement of Soft-teacher. Firstly, using mixed data enhancement to bring perturbation to the dataset and further utilize unlabeled images to enhance the effect of consistent training; then Swin-Transformer will be used as the backbone network to reduce the information loss caused by pooling, using shifted feature pyramids to do Multi-scale training, it can use the same label to supervise inputs of different sizes to obtain high-quality pseudo-labeling. Experiments on the NEU-DET dataset show that the method achieves good results in semi-supervised strip defect detection, and obtains 68.1% AP50 detection accuracy with 20% of labeled data, which is high enough to meet the needs of preliminary defect detection and labeling.
Surface defect detection; Deep learning; Semi-Supervised; Object detection
Heng Xiao, Changwei Zhao, Zhiyong Zhang. A Semi-Supervised Method for Steel Surface Defect Detection Based on Soft-teacher. Academic Journal of Computing & Information Science (2023), Vol. 6, Issue 3: 11-19. https://doi.org/10.25236/AJCIS.2023.060302.
[1] Liu, Y., Xu, K., & Xu, J. (2019). An improved MB-LBP defect recognition approach for the surface of steel plates. Applied Sciences, 9(20), 4222.
[2] Fu, G., Sun, P., Zhu, W., Yang, J., Cao, Y., Yang, M. Y., & Cao, Y. (2019). A deep-learning-based approach for fast and robust steel surface defects classification. Optics and Lasers in Engineering, 121, 397-405.
[3] Cheng, X., & Yu, J. (2020). RetinaNet with difference channel attention and adaptively spatial feature fusion for steel surface defect detection. IEEE Transactions on Instrumentation and Measurement, 70, 1-11.
[4] Zhao, C., Chen, Y., & Ma, J. (2018, October). Fabric defect detection algorithm based on MFS and SVM. In 2018 International Conference on Image and Video Processing, and Artificial Intelligence (Vol. 10836, pp. 77-82). SPIE.
[5] Liu, W., Anguelov, D., Erhan, D., Szegedy, C., Reed, S., Fu, C. Y., & Berg, A. C. (2016). Ssd: Single shot multibox detector. In Computer Vision–ECCV 2016: 14th European Conference, Amsterdam, The Netherlands, October 11–14, 2016, Proceedings, Part I 14 (pp. 21-37). Springer International Publishing.
[6] Redmon, J., Divvala, S., Girshick, R., & Farhadi, A. (2016). You only look once: Unified, real-time object detection. In Proceedings of the IEEE conference on computer vision and pattern recognition (pp. 779-788).
[7] Ren, S., He, K., Girshick, R., & Sun, J. (2015). Faster r-cnn: Towards real-time object detection with region proposal networks. Advances in neural information processing systems, 28.
[8] LI F G, YILIHAMU·Yaermaimaiti. Real-time Detection Model of Insulator Defect Based on Improved CenterNet [J]. Computer Science, 2022, 49(5): 84-91.
[9] Haselmann, M., & Gruber, D. P. (2019). Pixel-wise defect detection by CNNs without manually labeled training data. Applied Artificial Intelligence, 33(6), 548-566.
[10] He, Y., Song, K., Meng, Q., & Yan, Y. (2019). An end-to-end steel surface defect detection approach via fusing multiple hierarchical features. IEEE Transactions on Instrumentation and Measurement, 69(4), 1493-1504.
[11] Hao, X., Dong, T., & Zhang, D. (2021, November). A Highly Efficient Surface Defect Detection Approach for Hot Rolled Strip Steel Based on Deep Learning. In 2021 6th International Conference on Robotics and Automation Engineering (ICRAE) (pp. 318-322). IEEE.
[12] Yu, J., Cheng, X., & Li, Q. (2021). Surface defect detection of steel strips based on anchor-free network with channel attention and bidirectional feature fusion. IEEE Transactions on Instrumentation and Measurement, 71, 1-10.
[13] Li, X., Sun, Q., Liu, Y., Zhou, Q., Zheng, S., Chua, T. S., & Schiele, B. (2019). Learning to self-train for semi-supervised few-shot classification. Advances in neural information processing systems, 32.
[14] Xie, Q., Luong, M. T., Hovy, E., & Le, Q. V. (2020). Self-training with noisy student improves imagenet classification. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition (pp. 10687-10698).
[15] Miyato, T., Maeda, S. I., Koyama, M., & Ishii, S. (2018). Virtual adversarial training: a regularization method for supervised and semi-supervised learning. IEEE transactions on pattern analysis and machine intelligence, 41(8), 1979-1993.
[16] Sohn, K., Berthelot, D., Carlini, N., Zhang, Z., Zhang, H., Raffel, C. A., ... & Li, C. L. (2020). Fixmatch: Simplifying semi-supervised learning with consistency and confidence. Advances in neural information processing systems, 33, 596-608.
[17] Sohn, K., Zhang, Z., Li, C. L., Zhang, H., Lee, C. Y., & Pfister, T. (2020). A simple semi-supervised learning framework for object detection. arXiv preprint arXiv:2005.04757.
[18] Tarvainen, A., & Valpola, H. (2017). Mean teachers are better role models: Weight-averaged consistency targets improve semi-supervised deep learning results. Advances in neural information processing systems, 30.
[19] Liu, Y. C., Ma, C. Y., He, Z., Kuo, C. W., Chen, K., Zhang, P., ... & Vajda, P. (2021). Unbiased teacher for semi-supervised object detection. arXiv preprint arXiv:2102.09480.
[20] Xu, M., Zhang, Z., Hu, H., Wang, J., Wang, L., Wei, F., ... & Liu, Z. (2021). End-to-end semi-supervised object detection with soft teacher. In Proceedings of the IEEE/CVF International Conference on Computer Vision (pp. 3060-3069).
[21] Li, G., Li, X., Wang, Y., Wu, Y., Liang, D., & Zhang, S. (2022, November). Pseco: Pseudo labeling and consistency training for semi-supervised object detection. In Computer Vision–ECCV 2022: 17th European Conference, Tel Aviv, Israel, October 23–27, 2022, Proceedings, Part IX (pp. 457-472). Cham: Springer Nature Switzerland.
[22] Chen, B., Li, P., Chen, X., Wang, B., Zhang, L., & Hua, X. S. (2022). Dense learning based semi-supervised object detection. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (pp. 4815-4824).
[23] Liu, Z., Lin, Y., Cao, Y., Hu, H., Wei, Y., Zhang, Z. & Guo, B. (2021). Swin transformer: Hierarchical vision transformer using shifted windows. In Proceedings of the IEEE/CVF international conference on computer vision (pp. 10012-10022).