Zhou et al. (2026) MTU-Net: A Multiperspective Transformer U-Net With Water Index Channel-Spatial Attentions Embedding for Small Water Body Extraction From PlanetScope Imagery

Identification

• Journal: IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
• Year: 2026
• Date: 2026-01-01
• Authors: Pu Zhou, Yuyang Li, Yalan Wang, Xiang Li, Runsheng Ma, Yihang Zhang, Sisi Li, Yun Du, Xiaodong Li
• DOI: 10.1109/jstars.2026.3672155

Research Groups

[Information not available in the provided text snippet.]

Short Summary

This paper introduces MTU-Net, a novel deep learning architecture that integrates a multiperspective Transformer U-Net with water index channel-spatial attention mechanisms, designed for accurate and robust extraction of small water bodies from high-resolution PlanetScope imagery.

Objective

• To develop and evaluate MTU-Net, a new deep learning model incorporating a multiperspective Transformer U-Net and water index channel-spatial attention, for precise and efficient extraction of small water bodies from PlanetScope satellite imagery.

Study Configuration

• Spatial Scale: High resolution, suitable for small water body extraction (implied from PlanetScope imagery, typically 3-5 meters spatial resolution).
• Temporal Scale: [Information not available in the provided text snippet.]

Methodology and Data

• Models used: MTU-Net (Multiperspective Transformer U-Net with Water Index Channel-Spatial Attentions Embedding).
• Data sources: PlanetScope Imagery.

Main Results

• [Information not available in the provided text snippet.]

Contributions

• [Information not available in the provided text snippet.]

Funding

• [Information not available in the provided text snippet.]

Citation

@article{Zhou2026MTUNet,
  author = {Zhou, Pu and Li, Yuyang and Wang, Yalan and Li, Xiang and Ma, Runsheng and Zhang, Yihang and Li, Sisi and Du, Yun and Li, Xiaodong},
  title = {MTU-Net: A Multiperspective Transformer U-Net With Water Index Channel-Spatial Attentions Embedding for Small Water Body Extraction From PlanetScope Imagery},
  journal = {IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing},
  year = {2026},
  doi = {10.1109/jstars.2026.3672155},
  url = {https://doi.org/10.1109/jstars.2026.3672155}
}