Brazilian Vertical Traffic Signs and Lights Dataset
Description
The Brazilian Vertical Traffic Signs and Lights Dataset is a curated collection of annotated files containing images and information related to vertical traffic signs and lights in Brazil. This dataset is designed to facilitate research and development in the field of computer vision, machine learning, and intelligent transportation systems. Contents: 1. Image Data: - High-resolution images capturing various scenarios of Brazilian roadways. - Images feature a diverse range of vertical traffic signs and lights in different environmental conditions, lighting, and weather. 2. Annotation Files: - XML files containing detailed annotations for each image. - Annotations include bounding boxes around individual traffic signs and lights, along with corresponding class labels and other relevant attributes. 3. Classes: - Comprehensive taxonomy of Brazilian vertical traffic signs and lights. - Classes include regulatory signs, warning signs, guide signs, traffic lights, and other relevant categories as per Brazilian traffic regulations. 4. Metadata: - Additional metadata, such as geolocation, date, and time of image capture. - Information on road types, conditions, and any other contextual data that enhances the dataset's utility. Potential Applications: - Traffic sign and light recognition algorithms. - Intelligent transportation systems development. - Road safety and traffic management research. - Autonomous vehicle navigation and decision-making. Citation: When using this dataset, please cite the dataset as follows: \[Pierre, Monhel Maudoony (2023), “Brazilian Vertical Traffic Signs and Lights Dataset”, Mendeley Data, V2, doi: 10.17632/jbpsr4fvg9.2\] Acknowledgments: This dataset is compiled with data from some YouTube videos, and we acknowledge the contributions of the original data providers. Disclaimer: This dataset is provided for research purposes only. Users are encouraged to adhere to ethical guidelines and respect privacy regulations when using this data.
Files
Steps to reproduce
Creating the dataset of Brazilian vertical traffic signs and lights with annotated files involves several steps. Below is a general outline that you can follow: 1. Data Collection: - Photography or Video Recording: Capture high-resolution record videos of Brazilian roadways, ensuring diverse scenarios, lighting conditions, and weather. 2. Annotation: - Manual Annotation: Employ manual annotation tools to mark bounding boxes around each vertical traffic sign and light in the images or frames. - Class Labeling: Assign appropriate class labels to each annotated object, distinguishing between regulatory signs, warning signs, guide signs, and traffic lights. 3. Dataset Splitting: - Train-Test Split: Divide the dataset into training, testing and validation sets to evaluate model performance accurately. 4. Data Augmentation (Optional): - Apply data augmentation techniques to artificially increase the diversity of your dataset. This could include variations in rotation, scaling, brightness, and other transformations. 5. Data Formatting: - Organize the dataset in a structured format, such as a folder hierarchy, with separate directories for images and corresponding annotation files.