Add dataset card, paper link, and sample usage for Earth-Bench

README.md

@@ -1,3 +1,74 @@
----
-license: apache-2.0
----
+---
+license: apache-2.0
+task_categories:
+- image-text-to-text
+tags:
+- earth-observation
+- remote-sensing
+- multi-modal
+---
+
+# Earth-Bench
+
+[**Project Page**](https://opendatalab.github.io/Earth-Agent/) | [**Paper**](https://huggingface.co/papers/2509.23141) | [**GitHub**](https://github.com/opendatalab/Earth-Agent)
+
+Earth-Bench is a comprehensive benchmark designed to evaluate agentic frameworks in Earth Observation (EO). It consists of 248 expert-curated tasks with 13,729 images, spanning three primary modalities: Spectrum, Products, and RGB. The benchmark is equipped with a dual-level evaluation protocol that assesses both reasoning trajectories and final outcomes.
+
+The dataset was introduced as part of the paper [Earth-Agent: Unlocking the Full Landscape of Earth Observation with Agents](https://huggingface.co/papers/2509.23141).
+
+## Sample Usage
+
+### Download via huggingface-cli
+You can download the benchmark dataset using the `huggingface-cli`:
+
+```bash
+# Install huggingface-hub if not already installed
+pip install huggingface-hub
+
+# Download the dataset
+huggingface-cli download Sssunset/Earth-Bench --local-dir ./benchmark/data --repo-type dataset
+```
+
+### Download via Python
+Alternatively, you can use the `snapshot_download` function:
+
+```python
+from huggingface_hub import snapshot_download
+
+snapshot_download(
+    repo_id="Sssunset/Earth-Bench",
+    repo_type="dataset",
+    local_dir="./benchmark/data"
+)
+```
+
+## Dataset Structure
+
+The downloaded data follows this folder structure, organized by question:
+
+```
+data/
+├── question1/
+│   ├── image1
+│   ├── image2
+│   └── ...
+├── question2/
+│   ├── image1
+│   └── ...
+└── ...
+└── question248/
+    ├── image1
+    ├── image2
+    └── ...
+```
+
+## Citation
+
+```bibtex
+@article{feng2025earth,
+  title={Earth-Agent: Unlocking the Full Landscape of Earth Observation with Agents},
+  author={Feng, Peilin and Lv, Zhutao and Ye, Junyan and Wang, Xiaolei and Huo, Xinjie and Yu, Jinhua and Xu, Wanghan and Zhang, Wenlong and Bai, Lei and He, Conghui and others},
+  journal={arXiv preprint arXiv:2509.23141},
+  year={2025}
+}
+```