README.md

---
license: cc-by-sa-4.0
task_categories:
- image-classification
- object-detection
language:
- de
- en
tags:
- traffic-sign-recognition
- autonomous-driving
- computer-vision
- german-traffic-signs
- smart-city
- stvo
pretty_name: German Traffic SIGNs
size_categories:
- 10K<n<100K
thumbnail: teaser_image.png
---

# GTSIGN-220: A Crowd-Sourced, StVO-Aligned Dataset of German Traffic Signs

![Image Example for each Class](teaser_image.png)

**GTSIGN-220** (German Traffic SIGNs) is a comprehensive dataset containing **75,541 images** of traffic signs in Germany. 
It is specifically designed as a benchmark for fine-grained recognition, strictly aligned with the German Road Traffic Regulations (**StVO**).

The dataset is a derivative work of street-level imagery sourced from **Mapillary**. 
The individual image files of the signs are named so that the numbers before the first underscore correspond to the ID of the complete image from Mapillary. 
This allows the full image to be downloaded via the API.

---

## 📊 Dataset Summary

| Feature | Description |
| :--- | :--- |
| **Total Images** | 75,541 |
| **Classes** | 220 (StVO-aligned) |
| **Geography** | Germany |
| **Data Split** | Train, Validation, and Test sets |
| **Format** | Folders organized by class label |

---

## 📜 License & Attribution

This dataset is released under the **Creative Commons Attribution-ShareAlike 4.0 International ([CC BY-SA 4.0](https://creativecommons.org/licenses/by-sa/4.0/))** license.

### Derivative Work Notice
This dataset is a **derivative work** of street-level imagery originally sourced from **Mapillary**. In accordance with the "ShareAlike" (SA) provision of the original license, this dataset is released under the same CC BY-SA 4.0 terms.

### Required Attributions
To comply with the **Attribution (BY)** terms, you must credit both the dataset creators and the original source:

1.  **Creators:** Miriam Louise Carnot, ScaDS.AI (University of Leipzig).
2.  **Original Source:** Include the following statement:
    > "This work uses a dataset derived from street-level imagery from Mapillary (www.mapillary.com), originally contributed by 'Mapillary contributors' and licensed under CC BY-SA 4.0."

---

## 📖 Citation

If you use this dataset in your research, please cite the following paper:

```bibtex
@inproceedings{carnot_GTSIGN_2025,
  author    = {Carnot, Miriam Louise and Fastermann, Erik and Kunze, Jonas and Peukert, Eric and Ludwig, André and Franczyk, Bogdan},
  title     = {GTSIGN-220: A Crowd-Sourced, StVO-Aligned Benchmark for Fine-Grained German Traffic Sign Recognition},
  booktitle = {Intelligent Vehicles Symposium (IV)},
  year      = {2026}
}