RUGD (Unofficial Mirror)

Unofficial mirror of the RUGD dataset. I am not an author or owner of RUGD. All credit to the original creators.

The RUGD dataset contains outdoor, unstructured environments for autonomous navigation and visual perception research (e.g., semantic segmentation and scene understanding).

This repository re-hosts the original files as a convenience mirror for researchers who may have trouble accessing the official hosting due to connectivity or browser warnings. Note on the official site: the original download page has, at times, been served over plain HTTP or with configurations that some browsers flag as “Not secure.” This mirror is provided to make access easier; please prefer the official source when it works for you.

---

Attribution & Citation

If you use this dataset, please cite the original paper:

@inproceedings{RUGD2019IROS,
  author = {Wigness, Maggie and Eum, Sungmin and Rogers, John G and Han, David and Kwon, Heesung},
  title = {A RUGD Dataset for Autonomous Navigation and Visual Perception in Unstructured Outdoor Environments},
  booktitle = {International Conference on Intelligent Robots and Systems (IROS)},
  year = {2019}
}

Authors: Maggie Wigness, Sungmin Eum, John G. Rogers, David Han, Heesung Kwon Conference: IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 2019

---

What’s in this mirror

• Original RUGD archives/directories re-hosted for convenience.
• File names and structure preserved where possible.
• No modifications to the data itself.

If you’re an RUGD maintainer and would like this mirror updated or removed, please open a Discussion on this repo or contact me and I will promptly comply.

---

License & Usage

This mirror does not grant any license. Usage is governed by the original RUGD terms. Before using the data, review and comply with the upstream license/terms and any restrictions (e.g., research/non-commercial). If any license file was included upstream, it is mirrored here unchanged.

---

How to load (examples)

Option A — Use 🤗 imagefolder

If this repo is organized as folders of images (and optional labels), you can leverage the generic loader:

from datasets import load_dataset

# Replace "your-username/rugd" with the actual repo name
ds = load_dataset(
    "imagefolder",
    data_dir="hf://datasets/your-username/rugd",
    split="train"  # or "validation"/"test" if you provide splits
)

# Access a sample
sample = ds[0]
image = sample["image"]  # PIL.Image

If you provide segmentation masks in a parallel folder (e.g., images/ and annotations/), consider adding a small dataset script or a DatasetDict mapping so users get {"image": ..., "label": ...} pairs directly.

Option B — Stream raw files

If you just want to iterate over files as artifacts:

from huggingface_hub import hf_hub_download, list_repo_files

repo_id = "your-username/rugd"
files = list_repo_files(repo_id=repo_id, repo_type="dataset")

# Download a specific file (example)
path = hf_hub_download(repo_id=repo_id, repo_type="dataset", filename=files[0])

---

Intended tasks (non-exhaustive)

• Semantic Segmentation
• Scene Understanding / Traversability
• Perception for UGV/Off-road Robotics

For class definitions, color maps, and evaluation protocols, please refer to the original RUGD documentation/paper.

---

Provenance & Integrity

• Source: Official RUGD release (re-hosted).
• Files are mirrored “as-is.” If upstream checksums are available, you should verify them against the mirrored content.

---

Acknowledgements

All credit and thanks to the RUGD authors and their institutions for creating and releasing this dataset to the community.

---

Contact / Takedown

If you represent the original authors/rights holders and want this mirror changed or removed, please open a Discussion here or contact me directly and I’ll address it quickly.

---

Disclaimer: This repository is only a mirror to improve accessibility for researchers. I make no claim of authorship and no warranty about the content. Always consult and follow the original dataset’s terms.