SPLG-Mamba Prediction Maps

This package provides the saliency prediction maps of SPLG-Mamba on three ORSI-SOD benchmarks.

Contents

SPLG-Mamba_prediction_maps/
├── ORSSD/      # 200 prediction maps
├── EORSSD/     # 600 prediction maps
└── ORS-4199/   # 2199 prediction maps

The file names follow the corresponding test images and ground-truth masks in each dataset. All maps are stored as grayscale PNG saliency maps.

Usage

These maps are shared for reproducible comparison and academic research. They can be evaluated with standard SOD metrics such as MAE, S-measure, weighted F-measure, maximum F-measure, E-measure, and mean F-measure.

Citation

If these prediction maps are useful for your research, please cite our paper:

@misc{splg_mamba,
  title  = {SPLG-Mamba: Structure-Preserving Local-Global Mamba Network for Salient Object Detection in Optical Remote Sensing Images},
  author = {Xu, Yi and Hou, Ruichao and Ren, Tongwei and Wu, Gangshan},
  year   = {2026},
  note   = {Manuscript under review}
}