Abstract
Surface defect detection on steel components is crucial for quality control in polysilicon production. However, this task remains challenging due to tiny defect sizes, irregular geometries, complex backgrounds, and low contrast. To address these issues, we propose MSEOD-DDFusionNet (Multi-Scale and Effective Object-Detection Diffusion Fusion Network), a novel multi-scale diffusion-enhanced attention network. The network integrates four specialized modules: MTECAAttention (Multi-Scale Texture Enhancement Channel-Aware Attention) for lossless multi-scale feature fusion, ODConv (Omni-Dimensional Dynamic Convolution) for dynamic adaptation to irregular geometries, LMDP (Local Multi-Scale Discriminative Perception) for selective noise suppression and micro-defect amplification, and DDFusion (Diffusion-Driven Feature Fusion) for scene-aware noise modeling. Pruning further reduces computational complexity while improving accuracy. Extensive experiments on the specialized DDTE dataset and public benchmarks demonstrate state-of-the-art performance. Our model achieves 82.6% \(\hbox {mAP}_{50}\) and 61.6% \(\hbox {mAP}_{50-95}\) on DDTE, while maintaining a high inference speed of 193.5 FPS with only 8.46M parameters. It also shows excellent generalization across NEU-DET, GC10-DET, and cross-domain tasks, providing an efficient and accurate solution for industrial defect inspection.
Similar content being viewed by others
Data availability
The specific industrial dataset is subject to privacy restrictions and is not publicly available. To ensure reproducibility and enable further application, we provide the complete implementation, including code and pre-trained models, at: https://github.com/jiunian158/DDTE_Anonymous_Subset/blob/main/DDTE_Anonymous_Subset%20(3).zip
References
Wang, Z., Zhou, W. & Li, Y. MSAF-YOLO: An Efficient Multi-Scale Attention Fusion Network for high-precision steel surface defect detection. Measurement 118640, https://doi.org/10.1016/j.measurement.2025.118640 (2025).
Liu, R. et al. A lightweight model based on multi-scale feature fusion for ultrasonic welding surface defect detection. Eng. Appl. Artif. Intell. 161, 112208. https://doi.org/10.1016/j.engappai.2025.112208 (2025).
Wang, S. et al. Research on steel surface defect detection system based on YOLOv5s-SE-CA model and BEMD image enhancement. Nondestruct. Test. Eval. 1–20, https://doi.org/10.1080/10589759.2024.2393205 (2024).
Pan, Y. & Zhang, L. Dual attention deep learning network for automatic steel surface defect segmentation.. Comput.-Aided Civ. Inf. 37, 1468–1487. https://doi.org/10.1111/mice.12792 (2022).
Hosseini, S. M., Ebrahimi, A., Mosavi, M. R. & Shahhoseini, HSh. A novel hybrid CNN-CBAM-GRU method for intrusion detection in modern networks. Results Eng. 28, 107103. https://doi.org/10.1016/j.rineng.2025.107103 (2025).
Yeung, C.-C. & Lam, K.-M. Efficient fused-attention model for steel surface defect detection. IEEE Trans. Instrum. Meas. 71, 1–11. https://doi.org/10.1109/TIM.2022.3176239 (2022).
Ziadlou, G., Emami, S. & Asadi-Gangraj, E. Network configuration distributed production scheduling problem: A constraint programming approach. Comput. Ind. Eng. 188, 109916. https://doi.org/10.1016/j.cie.2024.109916 (2024).
Wen, J., Zheng, Y., Zhang, Y. & Yu, W. Enhanced dual-channel feature fusion approach for rolling bearing fault diagnosis. Nondestruct. Test. Eval. 40, 3309–3337. https://doi.org/10.1080/10589759.2025.2507761 (2025).
Ye, Z. & Yu, J. AKSNet: A novel convolutional neural network with adaptive kernel width and sparse regularization for machinery fault diagnosis. J. Manuf. Syst. 59, 467–480. https://doi.org/10.1016/j.jmsy.2021.03.022 (2021).
Xu, X., Li, X., Ming, W. & Chen, M. A novel multi-scale CNN and attention mechanism method with multi-sensor signal for remaining useful life prediction. Comput. Ind. Eng. 169, 108204. https://doi.org/10.1016/j.cie.2022.108204 (2022).
Zhou, H. et al. Ast-gnn: An attention-based spatio-temporal graph neural network for interaction-aware pedestrian trajectory prediction. Neurocomputing 445, 298–308. https://doi.org/10.1016/j.neucom.2021.03.024 (2021).
Rombach, R., Blattmann, A., Lorenz, D., Esser, P. & Ommer, B. High-resolution image synthesis with latent diffusion models. In Proc. IEEE/CVF Conf. Comput. Vis. Pattern Recognit. 10684–10695, https://doi.org/10.48550/arXiv.2112.10752 (2022).
Zhai, C., Wang, L. & Yuan, J. New Fusion Network with Dual-Branch Encoder and Triple-Branch Decoder for Remote Sensing Image Change Detection. Appl. Sci. 13, 6167. https://doi.org/10.3390/app13106167 (2023).
Fang, T. et al. Human-Guided Data Augmentation via Diffusion Model for Surface Defect Recognition Under Limited Data. IEEE Trans. Instrum. Meas. 74, 1–16. https://doi.org/10.1109/TIM.2025.3541684 (2025).
Lin, Z., Li, Z., Yu, J., Hu, M. & Wang, X. FFDDNet: Flexible focused defect detection network. IEEE Trans. Instrum. Meas. 74, 5019812. https://doi.org/10.1109/TIM.2025.3551459 (2025).
Wang, D., Shang, K., Wu, H. & Wang, C. Decoupled R-CNN: Sensitivity-specific detector for higher accurate localization. IEEE Trans. Circuits Syst. Video Technol. 32, 6324–6336. https://doi.org/10.1109/TCSVT.2022.3167114 (2022).
Fang, R. et al. FeatAug-DETR: Enriching one-to-many matching for DETRs with feature augmentation. IEEE Trans. Pattern Anal. Mach. Intell. 46, 6402–6415. https://doi.org/10.1109/TPAMI.2024.3381961 (2024).
Yang, K. et al. DETA: A point-based tracker with deformable transformer and task-aligned learning. IEEE Trans. Multimed. 25, 7545–7558. https://doi.org/10.1109/TMM.2022.3223213 (2023).
Zhang, M. et al. Oriented-DINO: Angle decoupling prediction and consistency optimizing for oriented detection transformer. IEEE Trans. Geosci. Remote Sens. 62, 5638315. https://doi.org/10.1109/TGRS.2024.3450200 (2024).
Li, Q., Shao, Y., Li, L., Li, J. & Hao, H. Weak surface defect detection for production-line plastic bottles with multi-view imaging system and LFF YOLO. Opt. Laser Eng. 181, 108369. https://doi.org/10.1016/j.optlaseng.2024.108369 (2024).
Zhu, C., Sun, Y., Zhang, H., Yuan, S. & Zhang, H. LE-YOLOv5: A lightweight and efficient neural network for steel surface defect detection. IEEE Access 12, 195242–195255. https://doi.org/10.1109/ACCESS.2024.3519161 (2024).
Liu, M., Chen, Y., Xie, J., He, L. & Zhang, Y. LF-YOLO: A lighter and faster yolo for weld defect detection of X-ray image. IEEE Sens. J. 23, 7430–7439. https://doi.org/10.1109/JSEN.2023.3247006 (2023).
Sun, J., Shen, X., Huang, H., Wang, Q. & Zhao, H. RT-SLAM: A Real-Time Visual SLAM System Integrating Enhanced RT-DETR and Optical Flow Techniques. IEEE Internet Things J. 12, 12803–12814. https://doi.org/10.1109/JIOT.2024.3522490 (2024).
Funding
This research is funded by the Autonomous Region Science and Technology Plan Project “Research and Application of Artificial Intelligence Technology for Silicon-Based New Material Manufacturing” (Project No. 2023B01033).
Author information
Authors and Affiliations
Contributions
Yiwei Duan contributed to the study conception and design. Material preparation, data collection, and analysis were performed by Others. The first draft of the manuscript was written by Yiwei Duan; all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.
Corresponding author
Ethics declarations
Competing interests
Each named author has substantially contributed to conducting the underlying research and drafting this manuscript. Additionally, to the best of our knowledge, the named authors have no conflict of interest, financial or otherwise.
Additional information
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.
About this article
Cite this article
Duan, Y., He, L., Wang, Z. et al. Multiscale diffusion-enhanced attention network for steel surface defect detection in Polysilicon Production. Sci Rep (2026). https://doi.org/10.1038/s41598-026-35913-8
Received:
Accepted:
Published:
DOI: https://doi.org/10.1038/s41598-026-35913-8
