Project-AgML/UAV_SaSeg_spartina_segmentation

Uav Saseg Spartina Segmentation

UAV-SaSeg is a high-resolution UAV dataset for mapping the invasive salt-marsh grass Spartina alterniflora (smooth cordgrass) on the coastal tidal flats of Sanmen County, Zhejiang, China, where it competes with native vegetation and takes over aquaculture areas. The tiles were cut from an RTK-georeferenced RGB orthomosaic with a 2 cm ground sample distance, captured in October 2025 when the plant was mature, so the set includes dense stands, small early-stage patches, and confusing backgrounds such as mudflats, water ripples, and native vegetation. Each 1024 x 1024 tile has a binary pixel mask (0 = background, 1 = Spartina alterniflora) drawn in LabelMe and ArcGIS and checked by experts. The dataset contains 358 images with pixel-level mask annotations.

This dataset is indexed on https://project-agml.github.io/ as part of the AgML python library.

Citation

@article{chen2026fine,
  title={Fine-Scale Spartina alterniflora Mapping Using Advanced Deep Learning and High-Resolution UAV Imagery},
  author={Chen, Zegang and Tang, Yixing and Hu, Yulong and Chu, Bin and Long, Zhong and Xie, Dong and Zhang, Yunfei and Wang, Zhipan},
  journal={Remote Sensing},
  volume={18},
  pages={2375},
  year={2026},
  publisher={MDPI}
}

The dataset itself can be cited as:

Tang. (2026). UAV-SaSeg [Dataset]. Zenodo. https://doi.org/10.5281/ZENODO.18697416

This dataset was reformatted from its original format to match HuggingFace standards.