The historical development of road infrastructure has relied on the human visual system as the primary method for perceiving and controlling vehicles. In order for an autonomous driving system to operate safely in existing road environments, it is crucial for it to possess visual recognition capabilities equivalent to those of humans. Visual classification, which includes tasks like object detection and semantic segmentation, plays a critical role in the visual perception of autonomous vehicles, particularly for accurately identifying road signs.

To facilitate research in this area, the German Traffic Sign Recognition Benchmark (GTSRB) dataset has been created. This dataset consists of images of traffic signs captured from German roads, with each image labeled according to its corresponding class. The dataset focuses on single-image, multi-class classification challenges and does not include any temporal information from the original video footage.

Here are some details about the GTSRB dataset:

1.Each image in the dataset has a resolution of 32 x 32 pixels and is represented in RGB format with three color channels. The pixel values are stored as unsigned 8-bit integers, providing a total of 256 possible values for each pixel.

2.The dataset comprises a total of 43 distinct classes or labels, based on the design or meaning of the traffic signs.

3.The training set consists of 34,799 images, each associated with its corresponding label.

4.The validation set contains 4,410 images, also accompanied by their respective labels.

5.Lastly, the test set comprises 12,630 images, each labeled with its corresponding class.

In CSV file,there are two columns. One is ClassID (represents the class ID of traffic signs) and another is SignName (represents the name of traffic sign).

GTSRB Traffic Sign Detection

## Overview

GTSRB Traffic Sign Detection is a dataset for object detection tasks - it contains Trafficsign annotations for 10,000 images.

## Getting Started

You can download this dataset for use within your own projects, or fork it into a workspace on Roboflow to create your own model.

  ## License

  This dataset is available under the [CC BY 4.0 license](https://creativecommons.org/licenses/CC BY 4.0).

Context

The German Traffic Sign Benchmark is a multi-class, single-image classification challenge held at the International Joint Conference on Neural Networks (IJCNN) 2011. We cordially invite researchers from relevant fields to participate: The competition is designed to allow for participation without special domain knowledge. Our benchmark has the following properties:

• Single-image, multi-class classification problem
• More than 40 classes
• More than 50,000 images in total
• Large, lifelike database

Acknowledgements

INI Benchmark Website

The GTSRB dataset consists of images of German traffic signs, utilized in the paper for evaluating the classification error and the impact of alignment on recognition.

Vision GTSRB

## Overview

Vision GTSRB is a dataset for object detection tasks - it contains Traffic Signs annotations for 875 images.

## Getting Started

You can download this dataset for use within your own projects, or fork it into a workspace on Roboflow to create your own model.

  ## License

  This dataset is available under the [CC BY 4.0 license](https://creativecommons.org/licenses/CC BY 4.0).

Dataset

This dataset was created by ibrahim karatas

Contents

Here are a few use cases for this project:

1. Autonomous Vehicle Navigation: The model can be used to identify traffic signs in real-time to assist autonomous vehicles in their navigation and decision-making processes. It could enable self-driving vehicles to follow speed limits, yield to signs, and take the appropriate action at intersections, thus improving safety.

2. Traffic Monitoring Systems: It can be integrated into traffic monitoring systems to track compliance with traffic rules and identify infractions. For instance, it can detect when vehicles exceed the speed limit or fail to yield where required and report such cases for appropriate actions.

3. Driving Assistant Applications: A driving assistant app can implement this model to provide audio and visual alerts to drivers about upcoming traffic signs, helping to increase driver's awareness and reduce the chances of traffic infractions and accidents.

4. Traffic Sign Inventory: Municipalities or highway authorities could use this model to keep an inventory of traffic signs across a city or highway, ensuring the signs are in the right condition and position.

5. Virtual Reality (VR) Driving Simulations: This model can be used to improve VR driving simulations, providing a more realistic environment by identifying and reacting to virtual traffic signs.

This dataset accompanies the article “Attention to Detail: A Conditional Multi-Head Transformer for Traffic Sign Recognition.” It includes pre-processed image data used for model training and testing. This dataset is derived from the publicly available German Traffic Sign Recognition Benchmark (GTSRB) dataset (Stallkamp et al., 2012). The current release includes a curated and preprocessed subset used for Transformer-based traffic sign recognition experiments. The original GTSRB dataset is available from Kaggle at https://www.kaggle.com/datasets/meowmeowmeowmeowmeow/gtsrb-german-traffic-sign.

Description

This dataset is pre-processed from the German Traffic Sign Recognition Benchmark (GTSRB) dataset. The original data include street-view photos of 43 different German traffic signs. As the photo images capture scenes larger than the traffic signs, the original dataset also provides coordinates to locate traffic sign within each image. This dataset is the result of cropping images with these provided coordinates, and the process is described in the attached jupyter notebook (german-traffic-signs-preprocessing.ipynb).

Acknowledgements

This dataset is built from the data made available at: https://www.kaggle.com/meowmeowmeowmeowmeow/gtsrb-german-traffic-sign. The original source is INI Benchmark (http://benchmark.ini.rub.de/?section=gtsrb) and its website provides detailed description on the dataset.

claudiogsc/GTSRB dataset hosted on Hugging Face and contributed by the HF Datasets community

GTSRB

## Overview

GTSRB is a dataset for object detection tasks - it contains Sign annotations for 1,461 images.

## Getting Started

You can download this dataset for use within your own projects, or fork it into a workspace on Roboflow to create your own model.

  ## License

  This dataset is available under the [CC BY 4.0 license](https://creativecommons.org/licenses/CC BY 4.0).

Description of what it is and how I did it

The data was collected manually (5-40 images per class) and afterwards augmented using 5 different functions at random:

RandomFog - addds some fog to the image; RandomSnow - adds the effect of snow on the image; Rotate + RandomScale - rotates the image to left or right by at maximum 30 degrees and changes image's quality; Affine - geometrically changes the image's shape to make it look unique; Perspective - changes the perspective of the image, making it unique as well.

One image got affected by 5 transforms for 7 times, giving 35 images as an output from only one. Then the process was repeated for the rest of the images.

TEST

Test folder include complitely mixed order of images (linked with Test_data.csv), all the images were extracted (completely removed) from train folder and augmented so as to avoid data leakage.

TEST_DATA.CSV

Test_data.csv is used after training to evaluate model's performance on a test set in Test folder. The Test_data.csv file include all images from Test folder in a mixed order.

META

The Meta folder consists of pattern images for each class.

TRAIN

The train folder consists of 205 folder enumerated from 0 to 204 making it easier to loop through them while image preprocessing. Each of the nested folders contain images of one traffic sign (cropped and augmented images from google images).

P.S. A friend of mine helped me with manually collecting the data and gave some usefull advices, you can find here -> A helpfull teammate

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).

For details on how the dataset was… See the full description on the dataset page: https://huggingface.co/datasets/SomeBottle/GTSRB_224x224_balanced.

Daxuxu36/ViT-GTSRB dataset hosted on Hugging Face and contributed by the HF Datasets community

The German Traffic Sign Benchmark is a multi-class, single-image classification challenge held at the International Joint Conference on Neural Networks (IJCNN) 2011. We cordially invite researchers from relevant fields to participate: The competition is designed to allow for participation without special domain knowledge.

Pre-calculated features

To allow scientists without a background in image processing to participate, we several provide pre-calculated feature sets. Each feature set contains the same directory structure as the training image set. For details on the parameters of the feature algorithm, please have a look at the file Feature_description.txt which is part of each archive file.

# HOG features

The file contains three sets of differently configured HOG features (Histograms of Oriented Gradients). The sets contain feature vectors of length 1568, 1568, and 2916 respectively. The features were calculated using the source code from http://pascal.inrialpes.fr/soft/olt/. For detailed information on HOG, we refer to

N. Dalal and B. Triggs. Histograms of Oriented Gradients for Human Detection. IEEE Conference on Computer Vision and Pattern Recognition, pages 886-893, 2005

Dataset

This dataset was created by HARIHARAN S 18ITR033

Contents

GTSRB dataset for Mamba-Transformer in Mamba-Transformer

This dataset consists of 15 classes of road traffic signs, each representing a specific road instruction or warning. The dataset is primarily intended for image classification tasks such as training deep learning models for autonomous driving systems, traffic sign recognition modules, and intelligent transport systems.

Key Characteristics

Total Images: 36.8k+ Number of Classes: 15 Type: Single-label classification (each image belongs to one class).

Potential Use Cases:

• Benchmarking traffic sign recognition algorithms
• Data augmentation and imbalance handling studies
• Fine-tuning of pre-trained image classification models (e.g. ResNet, Inception)

Datasets Used:

1) GTSRB - German Traffic Sign Recognition Benchmark Link: https://www.kaggle.com/datasets/meowmeowmeowmeowmeow/gtsrb-german-traffic-sign/code

The German Traffic Sign Benchmark is a multi-class, single-image classification challenge held at the International Joint Conference on Neural Networks (IJCNN) 2011.

2) Self-Driving Cars Computer Vision Project Link: https://universe.roboflow.com/selfdriving-car-qtywx/self-driving-cars-lfjou/dataset/6
Acquiring speed limit 20, speed limit 60, speed limit 100, stop sign

3) Traffic Sign Dataset Link: https://www.kaggle.com/datasets/ahemateja19bec1025/traffic-sign-dataset-classification/data?select=traffic_Data
Acquiring Turn left, pedestrian, one way, speed limit 60

This contains the 10 datasets used in the Visual Domain Decathlon, part of the PASCAL in Detail Workshop Challenge (CVPR 2017). The goal of this challenge is to solve simultaneously ten image classification problems representative of very different visual domains.

Some of the datasets included here are also available as separate datasets in TFDS. However, notice that images were preprocessed for the Visual Domain Decathlon (resized isotropically to have a shorter size of 72 pixels) and might have different train/validation/test splits. Here we use the official splits for the competition.

To use this dataset:

import tensorflow_datasets as tfds

ds = tfds.load('visual_domain_decathlon', split='train')
for ex in ds.take(4):
 print(ex)

See the guide for more informations on tensorflow_datasets.

https://storage.googleapis.com/tfds-data/visualization/fig/visual_domain_decathlon-aircraft-1.2.0.png" alt="Visualization" width="500px">