---
license: cc0-1.0
task_categories:
- image-classification
language:
- en
tags:
- traffic
- sign
pretty_name: Balanced GTSRB with 224x224 images
size_categories:
- 10K<n<100K
---

# Balanced GTSRB (224x224)  

This is a balanced GTSRB dataset containing 43 classes, with **1,000 training samples per class** and the same number of test samples as in the original dataset.  
All images have been **resized to 224×224** using interpolation and padding, maintaining aspect ratio.  

* For classes with fewer than 1,000 training samples, data augmentation was used to supplement the dataset (Note: without any flip transforms, thanks to [this post](https://medium.com/%40wolfapple/traffic-sign-recognition-2b0c3835e104)).

For details on how the dataset was generated, you can refer to the `flow.ipynb` notebook in the repository.  

* The dataset processing was assisted by AI coding.

## Structure

`gtsrb_224_balanced.zip`:  

```
GTSRB_224_balanced
├── test
│   ├── 00
│   │   ├── 00243.jpeg
│   ... ...
│       └── 12502.jpeg
└── train
    ├── 00
    │   ├── 00000_00000.jpeg
    │   ...
    ├── 01
    ...
```

## License

According to the [official site](https://benchmark.ini.rub.de/gtsrb_dataset.html), the dataset is free to use.   

And its copy was released with **CC0 License** in the [Kaggle repo](https://www.kaggle.com/datasets/meowmeowmeowmeowmeow/gtsrb-german-traffic-sign).  

## Citation

```bibtex
@inproceedings{Stallkamp-IJCNN-2011,
    author = {Johannes Stallkamp and Marc Schlipsing and Jan Salmen and Christian Igel},
    booktitle = {IEEE International Joint Conference on Neural Networks},
    title = {The {G}erman {T}raffic {S}ign {R}ecognition {B}enchmark: A multi-class classification competition},
    year = {2011},
    pages = {1453--1460}
}
```