Add model card for Co2S

This PR adds a model card for the Co2S model, including metadata, links to the paper, project page, and official GitHub repository. It also provides a brief description of the framework and its citation information.

README.md

@@ -0,0 +1,31 @@
+---
+pipeline_tag: image-segmentation
+tags:
+- remote-sensing
+- semi-supervised-learning
+---
+
+# Toward Stable Semi-Supervised Remote Sensing Segmentation via Co-Guidance and Co-Fusion
+
+This repository contains the weights for **Co2S**, a stable semi-supervised remote sensing (RS) image semantic segmentation framework.
+
+- **Paper:** [Toward Stable Semi-Supervised Remote Sensing Segmentation via Co-Guidance and Co-Fusion](https://huggingface.co/papers/2512.23035)
+- **Project Page:** [https://xavierjiezou.github.io/Co2S/](https://xavierjiezou.github.io/Co2S/)
+- **Code:** [https://github.com/XavierJiezou/co2s](https://github.com/XavierJiezou/co2s)
+
+## Description
+
+Co2S addresses the challenge of pseudo-label drift in semi-supervised remote sensing segmentation by synergistically fusing priors from vision-language models (CLIP) and self-supervised models (DINOv3). The framework utilizes a heterogeneous dual-student architecture, an explicit-implicit semantic co-guidance mechanism, and a global-local feature collaborative fusion strategy to achieve high-precision segmentation results across diverse RS datasets.
+
+## Citation
+
+If you find this work useful, please consider citing the original paper:
+
+```bibtex
+@article{zou2025toward,
+  title={Toward Stable Semi-Supervised Remote Sensing Segmentation via Co-Guidance and Co-Fusion},
+  author={Zou, Jie and others},
+  journal={arXiv preprint arXiv:2512.23035},
+  year={2025}
+}
+```