PC-SAM: Patch-Constrained Fine-Grained Interactive Road Segmentation in High-Resolution Remote Sensing Images

PC-SAM is a fine-grained interactive road segmentation model for high-resolution remote sensing images, it supports both automatic road segmentation and interactive segmentation refinement. By using point prompts, users can correct segmentation errors locally and obtain more accurate road segmentation masks.

Model Details

Model Description

• License: MIT
• Finetuned from model: SAM

Model Sources

• Repository: PC-SAM Github
• Paper: PC-SAM: Patch-Constrained Fine-Grained Interactive Road Segmentation in High-Resolution Remote Sensing Images

Uses/How to Get Started with the Model

Please refer to the GitHub page of PC-SAM.

Citation

@article{lv2026pc,
  title={PC-SAM: Patch-Constrained Fine-Grained Interactive Road Segmentation in High-Resolution Remote Sensing Images},
  author={Lv, Chengcheng and Li, Rushi and Wu, Mincheng and Shi, Xiufang and Wen, Zhenyu and He, Shibo},
  journal={arXiv preprint arXiv:2604.00495},
  year={2026}
}