EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

D2FLS-Net: Dual-Stage DEM-guided Fusion Transformer for landslide segmentation

Chengwei Zhao iD, Long Li, Yubo Wang, Xuqing Li iD, Chong Xu iD, Yubin Song, Dongsheng Ren, Cheng Xiao

PLOS One · 26.11.2025 · 20 (11) · e0337412

DOI: 10.1371/journal.pone.0337412 ↗

Kurz gefasst

Fachartikel in der Zeitschrift PLOS One, online erschienen am 26.11.2025. Verfasst von Chengwei Zhao und 7 weiteren Autor:innen. Beteiligte Einrichtung: School of Remote Sensing and Information Engineering, North China Institute of Aerospace Engineering, Langfang, China. Hauptthemen laut MEDLINE: Landslides, Remote Sensing Technology, Image Processing, Computer-Assisted. Die Arbeit wurde bisher 7-mal zitiert (Crossref) und nennt 39 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Hebei Provincial Central Government Guide Local Science and Technology Development Fund Project for Free Exploration Basic Research, Key Project of Hebei Provincial Department of Education, Doctoral Fund Project of North China Institute of Aerospace Engineering.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Landslide segmentation from remote sensing imagery is crucial for rapid disaster assessment and risk mitigation. Owing to the pronounced heterogeneity of landslide scales and the subtle visual contrast between some landslide bodies and their background, this task remains highly challenging. Although Transformers surpass convolutional neural networks in modeling long-range contextual dependencies, channel-level or feature-level fusion strategies provide only intermittent terrain cues, leading models to underutilize digital elevation model (DEM) information and to lack fine-grained adaptability to terrain variability. To address this, We propose a Swin-Transformer–based framework, Dual-Stage DEM-guided Fusion Transformer for landslide segmentation (D2FLS-Net), which embeds terrain features via two modules: (1) The Dual-Stage DEM-Guided Fusion (DSDF) module that injects DEM cues twice, where the early stage emphasizes DEM related discontinuities before feature extraction, and the late stage coordinates high-level RGB and DEM semantics through a cross-attention mechanism. (2) The Terrain-aware Pixel-wise Adaptive Context Enhancement (T-PACE) module that optimizes intermediate features using a DEM-gated, pixel-adaptive hybrid of multi-dilation atrous convolutions, enabling broader context aggregation within homogeneous landslide interiors and more precise discrimination at boundaries. We evaluate D2FLS-Net on the Bijie and Landslide4Sense 2022 datasets. On Bijie, the mean Intersection over Union (mIoU) reaches 88.77%, Recall 95.27%, and Precision 94.60%, exceeding the best competing model SegFormer by 7.96%, 7.90%, and 4.05%, respectively. On Landslide4Sense2022, mIoU 72.86%, Recall 82.55%, and Precision 93.30%, surpassing SegFormer by 7.06%, 6.56%, and 5.02%, respectively. Ablation studies indicate that DSDF primarily reduces missed detections of landslide traces, whereas T-PACE refines pixel level context selection. Injecting DEM at the Swin-1 and Swin-4 stages consistently outperforms other stage combinations. In summary, the model shows good detection performance and is suitable for fusing DEM and remote sensing imagery for landslide recognition.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

LandslidesRemote Sensing TechnologyImage Processing, Computer-AssistedNeural Networks, ComputerAlgorithms

Autor:innen und Institutionen

  • School of Remote Sensing and Information Engineering, North China Institute of Aerospace Engineering, Langfang, China.
  • State Key Laboratory of Ecological Safety and Sustainable Development in Arid Lands, Xinjiang Institute of Ecology and Geography, Chinese Academy of Sciences, Urumqi, China.
  • University of Chinese Academy of Sciences, Beijing, China.
  • National Institute of Natural Hazards, Ministry of Emergency Management of China, Beijing, China.
  • School of Electronic and Control Engineering, North China Institute of Aerospace Engineering, Langfang, China.
  • Hebei Provincial Institute of Geological Surveying and Mapping (Spatial Information Technology Application Research Center of Hebei Provincial Bureau of Geology and Mineral Exploration and Development), Langfang, China.

ORCID-Profile: Chengwei Zhao, Xuqing Li, Chong Xu

Förderung

  • Hebei Provincial Central Government Guide Local Science and Technology Development Fund Project for Free Exploration Basic Research
  • Key Project of Hebei Provincial Department of Education
  • Doctoral Fund Project of North China Institute of Aerospace Engineering

Zitieren

APA

Zhao, C., Li, L., Wang, Y., Li, X., Xu, C., Song, Y., Ren, D., & Xiao, C. (2025). D2FLS-Net: Dual-Stage DEM-guided Fusion Transformer for landslide segmentation. PLOS One, 20(11), e0337412. https://doi.org/10.1371/journal.pone.0337412

Vancouver

Zhao C, Li L, Wang Y, Li X, Xu C, Song Y, et al. D2FLS-Net: Dual-Stage DEM-guided Fusion Transformer for landslide segmentation. PLOS One. 2025;20(11):e0337412. doi:10.1371/journal.pone.0337412

BibTeX
@article{Zhao2025,
  author = {Zhao, Chengwei and Li, Long and Wang, Yubo and Li, Xuqing and Xu, Chong and Song, Yubin and Ren, Dongsheng and Xiao, Cheng},
  title = {D2FLS-Net: Dual-Stage DEM-guided Fusion Transformer for landslide segmentation},
  journal = {PLOS One},
  year = {2025},
  volume = {20},
  number = {11},
  pages = {e0337412},
  doi = {10.1371/journal.pone.0337412}
}
Kostenlos starten →