Applying Diffusion Models for Classification of Central Segregation in Steel Plates
Resumo
A classificação automática dos níveis de segregação em placas de lingotamento contínuo é crucial para garantir a qualidade do aço. A macrossegregação, caracterizada por variações químicas localizadas, representa um desafio significativo, especialmente na região central das placas. Este trabalho propõe uma abordagem baseada em redes neurais convolucionais, especificamente a EfficientNet-B0, com transferência de aprendizado, aliada à geração de dados sintéticos por meio de modelos de difusão (Stable Diffusion). Os resultados experimentais obtidos em um conjunto de treino de 150 imagens para treino e 30 para teste indicaram acurácia inicial de 80% sem nenhum aumento de dados, com uma melhora superior a 13% após a inclusão de dados sintéticos, além de ganhos em precisão e revocação. A proposta contribui para a automação da inspeção de falhas e se alinha aos princípios da Indústria 4.0.
Referências
Beckermann, C. (2002). Modelling of macrosegregation: Applications and future needs. International Materials Reviews, 47(5):243–261.
Chen, Z. and colleagues (2025). Stable diffusion models are secretly good at visual in-context learning. In International Conference on Learning Representations (ICLR). [link].
Costa, A. B. S., Nishiura, J. Y., Lopes, P. V., Verri, F. A. N., and Skoogh, A. (2024). Artificial data generation for smart manufacturing systems: Discrete event simulation, traceability, and process mining. In Encontro Nacional de Inteligência Artificial e Computacional (ENIAC), pages 719–730. SBC.
Fu, G., Zhang, Z., Le, W., Li, J., Zhu, Q., Niu, F., Chen, H., Sun, F., and Shen, Y. (2023). A multi-scale pooling convolutional neural network for accurate steel surface defects classification. Frontiers in Neurorobotics, 17:1096083.
Gao, Y., Gao, L., Li, X., and Yan, X. (2019). A semi-supervised convolutional neural network-based method for steel surface defect recognition. Robotics and Computer-Integrated Manufacturing, 61:101825.
Ho, J., Jain, A., and Abbeel, P. (2020). Denoising diffusion probabilistic models. In Advances in Neural Information Processing Systems, volume 33, pages 6840–6851.
Hu, Y., Tang, J., Xu, Y., Xu, R., and Huang, B. (2024). EL-DenseNet: A novel method for identifying the flame state of converter steelmaking based on dense convolutional neural networks. Signal, Image and Video Processing, 18:3445–3457.
Ibrahim, A. A. M. (2024). Transfer learning-based approach using new convolutional neural network classifier for steel surface defects classification. Scientific African, 23:e02066.
Li, H., Liu, Y., Liu, C., Pang, H., and Xu, K. (2025). A few-shot steel surface defect generation method based on diffusion models. Sensors, 25(10):3038. DOI: 10.3390/s25103038.
Lima, F. R. and Gomes, R. (2020). Conceitos e tecnologias da indústria 4.0: uma análise bibliométrica. Revista Brasileira de Inovação, 19:e0200023.
Ludwig, A., Wu, M., and Kharicha, A. (2015). On macrosegregation. Metallurgical and Materials Transactions A, 46:4854–4867.
Moore, J. (1984). Review of axial segregation in continuous cast steel. Continuous Casting, Iron and Steel Society, 3:11–20.
Müller-Franzes, G., Niehues, J. M., Khader, F., Arasteh, S. T., Haarburger, C., Kuhl, C., Wang, T., Han, T., Nolte, T., Nebelung, S., et al. (2023). A multimodal comparison of latent denoising diffusion probabilistic models and generative adversarial networks for medical image synthesis. Scientific Reports, 13(1):12098.
Nieto, P. J. G., García-Gonzalo, E., Álvarez Antón, J. C., Suárez, V. M. G., Bayón, R. M., and Martín, F. M. (2018). A comparison of several machine learning techniques for the centerline segregation prediction in continuous cast steel slabs and evaluation of its performance. Journal of Computational and Applied Mathematics, 330:877–895.
Ottoni, A. L. C., de Amorim, R. M., Novo, M. S., and Costa, D. B. (2023). Tuning of data augmentation hyperparameters in deep learning to building construction image classification with small datasets. International Journal of Machine Learning and Cybernetics, 14(1):171–186.
Queiroz, L. C. L. (2015). Comparison of edge detection techniques applied in the identification of centerline segregation on steel slabs. American Journal of Applied Sciences, 12(8):567–571.
Ruiz, N., Li, Y., Jampani, V., Pritch, Y., Rubinstein, M., and Aberman, K. (2022). Dreambooth: Fine tuning text-to-image diffusion models for subject-driven generation.
Ruiz, N., Li, Y., Jampani, V., Pritch, Y., Rubinstein, M., and Aberman, K. (2023). Dreambooth: Fine tuning text-to-image diffusion models for subject-driven generation. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 22500–22510.
Shorten, C. and Khoshgoftaar, T. M. (2019). A survey on image data augmentation for deep learning. Journal of Big Data, 6(1):60.
Tai, Y., Yang, K., Peng, T., Huang, Z., and Zhang, Z. (2021). Defect image sample generation with diffusion prior for steel surface defect recognition. Journal of LaTeX Class Files, 14(8):1–13. Available at: arXiv:2405.01872 [cs.CV].
Tan, M. and Le, Q. (2019). Efficientnet: Rethinking model scaling for convolutional neural networks. In Chaudhuri, K. and Salakhutdinov, R., editors, Proceedings of the 36th International Conference on Machine Learning, volume 97 of Proceedings of Machine Learning Research, pages 6105–6114. PMLR.
Zhang, C., Gao, W., Ma, C., Zhang, R., Yin, H., and Du, Y. (2024). Diffusion-based text-to-image generative model for predicting materials microstructure evolutions. SSRN Electronic Journal. Preprint, available at SSRN: [link].
Zhang, R. and Yang, J. (2023). State of the art in applications of machine learning in steelmaking process modeling. International Journal of Minerals, Metallurgy and Materials, 30(11):2055–2067.
Zou, L., Zhang, J., Liu, Q., Zeng, F., Chen, J., and Guan, M. (2019). Prediction of central carbon segregation in continuous casting billet using a regularized extreme learning machine model. Metals, 9(12):1312.
Chen, Z. and colleagues (2025). Stable diffusion models are secretly good at visual in-context learning. In International Conference on Learning Representations (ICLR). [link].
Costa, A. B. S., Nishiura, J. Y., Lopes, P. V., Verri, F. A. N., and Skoogh, A. (2024). Artificial data generation for smart manufacturing systems: Discrete event simulation, traceability, and process mining. In Encontro Nacional de Inteligência Artificial e Computacional (ENIAC), pages 719–730. SBC.
Fu, G., Zhang, Z., Le, W., Li, J., Zhu, Q., Niu, F., Chen, H., Sun, F., and Shen, Y. (2023). A multi-scale pooling convolutional neural network for accurate steel surface defects classification. Frontiers in Neurorobotics, 17:1096083.
Gao, Y., Gao, L., Li, X., and Yan, X. (2019). A semi-supervised convolutional neural network-based method for steel surface defect recognition. Robotics and Computer-Integrated Manufacturing, 61:101825.
Ho, J., Jain, A., and Abbeel, P. (2020). Denoising diffusion probabilistic models. In Advances in Neural Information Processing Systems, volume 33, pages 6840–6851.
Hu, Y., Tang, J., Xu, Y., Xu, R., and Huang, B. (2024). EL-DenseNet: A novel method for identifying the flame state of converter steelmaking based on dense convolutional neural networks. Signal, Image and Video Processing, 18:3445–3457.
Ibrahim, A. A. M. (2024). Transfer learning-based approach using new convolutional neural network classifier for steel surface defects classification. Scientific African, 23:e02066.
Li, H., Liu, Y., Liu, C., Pang, H., and Xu, K. (2025). A few-shot steel surface defect generation method based on diffusion models. Sensors, 25(10):3038. DOI: 10.3390/s25103038.
Lima, F. R. and Gomes, R. (2020). Conceitos e tecnologias da indústria 4.0: uma análise bibliométrica. Revista Brasileira de Inovação, 19:e0200023.
Ludwig, A., Wu, M., and Kharicha, A. (2015). On macrosegregation. Metallurgical and Materials Transactions A, 46:4854–4867.
Moore, J. (1984). Review of axial segregation in continuous cast steel. Continuous Casting, Iron and Steel Society, 3:11–20.
Müller-Franzes, G., Niehues, J. M., Khader, F., Arasteh, S. T., Haarburger, C., Kuhl, C., Wang, T., Han, T., Nolte, T., Nebelung, S., et al. (2023). A multimodal comparison of latent denoising diffusion probabilistic models and generative adversarial networks for medical image synthesis. Scientific Reports, 13(1):12098.
Nieto, P. J. G., García-Gonzalo, E., Álvarez Antón, J. C., Suárez, V. M. G., Bayón, R. M., and Martín, F. M. (2018). A comparison of several machine learning techniques for the centerline segregation prediction in continuous cast steel slabs and evaluation of its performance. Journal of Computational and Applied Mathematics, 330:877–895.
Ottoni, A. L. C., de Amorim, R. M., Novo, M. S., and Costa, D. B. (2023). Tuning of data augmentation hyperparameters in deep learning to building construction image classification with small datasets. International Journal of Machine Learning and Cybernetics, 14(1):171–186.
Queiroz, L. C. L. (2015). Comparison of edge detection techniques applied in the identification of centerline segregation on steel slabs. American Journal of Applied Sciences, 12(8):567–571.
Ruiz, N., Li, Y., Jampani, V., Pritch, Y., Rubinstein, M., and Aberman, K. (2022). Dreambooth: Fine tuning text-to-image diffusion models for subject-driven generation.
Ruiz, N., Li, Y., Jampani, V., Pritch, Y., Rubinstein, M., and Aberman, K. (2023). Dreambooth: Fine tuning text-to-image diffusion models for subject-driven generation. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 22500–22510.
Shorten, C. and Khoshgoftaar, T. M. (2019). A survey on image data augmentation for deep learning. Journal of Big Data, 6(1):60.
Tai, Y., Yang, K., Peng, T., Huang, Z., and Zhang, Z. (2021). Defect image sample generation with diffusion prior for steel surface defect recognition. Journal of LaTeX Class Files, 14(8):1–13. Available at: arXiv:2405.01872 [cs.CV].
Tan, M. and Le, Q. (2019). Efficientnet: Rethinking model scaling for convolutional neural networks. In Chaudhuri, K. and Salakhutdinov, R., editors, Proceedings of the 36th International Conference on Machine Learning, volume 97 of Proceedings of Machine Learning Research, pages 6105–6114. PMLR.
Zhang, C., Gao, W., Ma, C., Zhang, R., Yin, H., and Du, Y. (2024). Diffusion-based text-to-image generative model for predicting materials microstructure evolutions. SSRN Electronic Journal. Preprint, available at SSRN: [link].
Zhang, R. and Yang, J. (2023). State of the art in applications of machine learning in steelmaking process modeling. International Journal of Minerals, Metallurgy and Materials, 30(11):2055–2067.
Zou, L., Zhang, J., Liu, Q., Zeng, F., Chen, J., and Guan, M. (2019). Prediction of central carbon segregation in continuous casting billet using a regularized extreme learning machine model. Metals, 9(12):1312.
Publicado
29/09/2025
Como Citar
OLIVEIRA, Luciano M.; SILVA, Guilherme; NEGRÃO, Arthur; OTTONI, Andre L.; SILVA, Pedro.
Applying Diffusion Models for Classification of Central Segregation in Steel Plates. In: ENCONTRO NACIONAL DE INTELIGÊNCIA ARTIFICIAL E COMPUTACIONAL (ENIAC), 22. , 2025, Fortaleza/CE.
Anais [...].
Porto Alegre: Sociedade Brasileira de Computação,
2025
.
p. 1057-1068.
ISSN 2763-9061.
DOI: https://doi.org/10.5753/eniac.2025.14329.
