SCORE

<h2 align="center">SCORE: Scene Context Matters <br> in Open-Vocabulary Remote Sensing Instance Segmentation </h2> <div align="center"> <p> <a href="https://arxiv.org/abs/2507.12857"><img src="https://img.shields.io/badge/arXiv-SCORE-b31b1b.svg"></a> <a href="https://arxiv.org/pdf/2507.12857"><img src="https://img.shields.io/badge/PDF-8A2BE2"></a> </p> </div>

This repository contains code for our ICCV2025 Highlight✨ paper:

SCORE: Scene Context Matters in Open-Vocabulary Remote Sensing Instance Segmentation<br> Shiqi Huang, Shuting He, Huaiyuan Qin, Bihan Wen<br> ICCV 2025 (Highlight)

<div align="left"> <img src="imgs/teaser.png" width="70%" height="100%"/> </div><br/>

Framework

<div align="center"> <img src="imgs/framework.png" width="100%" height="100%"/> </div><br/>

Installation

Please see Installation Instructions.

Datasets

Download datasets for open-vocabulary remote sensing instance segmentation from Hugging Face <img src="https://huggingface.co/front/assets/huggingface_logo-noborder.svg" alt="Hugging Face Logo" width="20"/>.

Getting Started with SCORE

Please see Getting Started with Detectron2 for full usage.

Training

Download RS-CLIP ckpt from RemoteCLIP and put it under SCORE/RemoteCLIP/RemoteCLIP-ViT-L-14.pt.

python train_net.py --num-gpus 1 \
  --config-file configs/score_isaid_instances.yaml

python train_net.py --num-gpus 1 \
  --config-file configs/score_sior_instances.yaml

Inference

bash eval_isaid.sh

bash eval_sior.sh

Acknowledgement

This project is based on FC-CLIP. Many thanks to the authors for their great work!

BibTeX

Please consider to cite SCORE if it helps your research.

@inproceedings{SCORE,
  title={SCORE: Scene Context Matters in Open-Vocabulary Remote Sensing Instance Segmentation},
  author={Huang, Shiqi and He, Shuting and Qin, Huaiyuan and Wen, Bihan},
  booktitle={Proceedings of the IEEE/CVF International Conference on Computer Vision},
  pages={12559--12569},
  year={2025}
}