Hugging Face's logo

Datasets:
Pikachu
/
EuroRoads

Tasks:
Image Classification
Image Segmentation
Languages:
English
Size:
10K<n<100K
Tags:
roads
europe
driving
netherlands
belgium
germany
License:
Dataset card Files
xet
Community
EuroRoads / README.md
Pikachu's picture
Pikachu
Update README.md
d7db522 verified
preview code
|
raw
history blame contribute delete
3.95 kB
metadata
pretty_name: EuroRoads
license: cc-by-4.0
task_categories:
  - image-classification
  - image-segmentation
size_categories:
  - 10K<n<100K
language:
  - en
tags:
  - roads
  - europe
  - driving
  - netherlands
  - belgium
  - germany
configs:
  - config_name: default
    data_files:
      - split: train
        path: dataset/EuroRoads-13.7K.zip

🛣️ EuroRoads, a free to use European road dataset

A high-resolution collection of 13.7 thousand road scene images captured across the Netherlands (Limburg), Germany, and Belgium. The dataset focuses on realistic driving environments ranging from highways to small rural paths — designed for use in computer vision tasks such as segmentation, detection, or autonomous driving simulation. 10x crop halfres Three images side by side, below them an extreme 10x crop (for full res see assets folder)

📦 Dataset Summary

Property Description
Total Images 13,706
Resolution 3072 × 3072
Capture Device Xiaomi 14 Ultra / 15 Ultra (soon)
Frame Interval 1 second
Filtering Long traffic pauses removed
Geographic Focus Mainly Netherlands (Limburg), also Germany & Belgium
License CC-BY
Status Work in progress
Camera Mode Auto (may cause inconsistent motion blur)
Camera Style Mainly Leica Authentic

🗺️ Description

This dataset provides sharp and detailed images captured in various European environments and weather conditions. It covers a wide range of road types:

  • 🚗 Highways and intercity roads
  • 🚙 Regular city and suburban roads
  • 🚜 Narrow paths and village streets
  • 🌳 Paved roads through nature

All images were captured at 1-second intervals, with long idle pauses removed (red lights). The result is a mostly unfiltered, time-consistent dataset. All images have kept their original file names and metadata, although location information is excluded.

🧭 Geographic Context

  • Primary Region: Limburg, Netherlands
  • Additional Regions: Adjacent areas in Germany and Belgium
  • Environment Mix: Urban → rural → nature-paved roads

📷 Image Characteristics

  • Captured using high-end mobile sensors (Xiaomi 14/15 Ultra)
  • Auto mode exposure and focus
  • 1:1 aspect ratio optimized for dataset uniformity
  • Possible inconsistencies in motion blur or lighting due to automatic settings
  • No filtering or labeling beyond removal of long idle periods

🪪 License

Creative Commons Attribution (CC BY 4.0) You are free to share and adapt the data, provided appropriate credit is given.

🚧 Work in Progress

This dataset is actively being expanded and refined. Future releases may include:

  • Filtered subsets (e.g., motion blur score, weather condition, time of day)
  • Semantic annotations or segmentation masks

📬 Citation

If you use this dataset in your work, please cite it as:

@dataset{road_scenes_limburg_de_be_2025,
  title        = {EuroRoads (Limburg–DE–BE) Dataset},
  author       = {Randy Hübner},
  year         = {2025},
  license      = {CC-BY-4.0},
  url          = {https://huggingface.co/datasets/Pikachu/EuroRoads}
}

license: cc-by-4.0

configs:

  • config_name: default data_files:
    • split: train path: dataset/EuroRoads-13.7K.zip