EUVIMEDEUVIMED entdecken
Crossref

ISF-UNet: A lightweight 3D seismic fault segmentation network with inception depthwise convolution and channel attention fusion

Lijie Cui, Yawen Huang, Yuxi Niu, Ye Tao, Song Bai, Jiaqi Zhao, Paerhati Piluolan

Interpretation · 01.11.2026 · 14 (4) · T411-T424

DOI: 10.1190/int-2025-1008 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Interpretation, online erschienen am 03.11.2026. Verfasst von Lijie Cui und 6 weiteren Autor:innen. Die Arbeit nennt 26 Literaturangaben. Gefördert von: the Research Foundation of China University of Petroleum-Beijing at Karamay.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Accurate detection of seismic faults is critical for geologic exploration, reservoir characterization, and hazard assessment, yet traditional approaches remain vulnerable to noise and often rely heavily on manual interpretation. To address these limitations, a lightweight 3D seismic fault segmentation network, ISF-UNet, was developed with a central innovation in the ISF module, which integrated an inception depthwise convolution submodule and a squeeze-and-excite fusion submodule. The ISF module used a multibranch parallel structure with a shared channel attention mechanism to capture multiscale spatial features while maintaining efficiency with minimal parameter overhead. A composite Dice–Focal loss function was further incorporated to mitigate class imbalance and enhance segmentation performance. The network was trained with multiple data augmentation strategies to improve generalization. Evaluation on synthetic and real seismic data sets, including F3, Kerry-3D, and field data from an oilfield in the Junggar Basin, China, demonstrated that ISF-UNet achieved superior recognition accuracy and structural continuity of faults compared with conventional architectures. The lightweight design not only ensured robust feature extraction but also provided practical advantages for deployment in resource-constrained environments. These results highlighted the effectiveness of integrating multibranch convolutional operations with channel attention and tailored loss functions, offering a reliable and scalable solution for seismic fault detection.

Förderung

  • the Research Foundation of China University of Petroleum-Beijing at Karamay

Zitieren

APA

Cui, L., Huang, Y., Niu, Y., Tao, Y., Bai, S., Zhao, J., & Piluolan, P. (2026). ISF-UNet: A lightweight 3D seismic fault segmentation network with inception depthwise convolution and channel attention fusion. Interpretation, 14(4), T411-T424. https://doi.org/10.1190/int-2025-1008

Vancouver

Cui L, Huang Y, Niu Y, Tao Y, Bai S, Zhao J, et al. ISF-UNet: A lightweight 3D seismic fault segmentation network with inception depthwise convolution and channel attention fusion. Interpretation. 2026;14(4):T411-T424. doi:10.1190/int-2025-1008

BibTeX
@article{Cui2026,
  author = {Cui, Lijie and Huang, Yawen and Niu, Yuxi and Tao, Ye and Bai, Song and Zhao, Jiaqi and Piluolan, Paerhati},
  title = {ISF-UNet: A lightweight 3D seismic fault segmentation network with inception depthwise convolution and channel attention fusion},
  journal = {Interpretation},
  year = {2026},
  volume = {14},
  number = {4},
  pages = {T411--T424},
  doi = {10.1190/int-2025-1008}
}
Kostenlos starten →