Dark Mode
YOLO-ready Traffic Signal Dataset
Data Science and Analytics
Tags and Keywords
Trusted By
"No reviews yet"
Free
About
This product offers annotated image data specifically designed for Traffic Sign Localisation and Detection models 🚦. The collection was assembled from multiple sources to provide a suitable training set. A key usage instruction is to refrain from applying rotational augmentation to the dataset. Rotational changes can corrupt the integrity of the signs, for example, confusing '80' with '08' or switching turn directions, which will consequently lessen detection accuracy. Maximum model accuracy is achieved when using only the original dataset files.
Columns
The dataset includes a companion metadata file,
labels_YOLO_Traffic_Sign.csv (749 B), which details the necessary labelling information.- ClassID: This is a numeric identifier for each traffic sign type. This column contains 39 unique values, ranging from 1 up to a maximum of 39, with a mean value of 20.
- Description: Provides the textual label corresponding to the ClassID. This column also holds 39 distinct unique values, with "Pedestrian" being listed among the most common entries.
Distribution
The core of this data product is an image collection featuring around 2,500 images suitable for detection tasks. The associated metadata file is supplied in CSV format. Code handling assistance is available within an associated Notebook titled "Official".
Usage
This data is highly suitable for machine learning engineers and developers working on intelligent transportation systems and computer vision applications. Ideal uses include:
- Training models for precise localisation and identification of various road signs.
- Developing Multiclass Classification algorithms using Keras and image data.
- Educational or beginner projects focused on object detection, offering practical experience with YOLO annotation formats.
Coverage
The scope covers 39 unique traffic sign classifications, as detailed in the metadata. Specific geographic or time boundaries are not defined, as the collection was gathered from different sources. The expected frequency for updates to this dataset is monthly.
License
CC0: Public Domain
Who Can Use It
This dataset is designed for use by a variety of users interested in object detection and image analysis:
- Data Scientists: For developing and benchmarking traffic sign recognition algorithms.
- Students/Beginners: To gain practical experience with image annotation and building foundational computer vision models.
- Automotive Engineers: For integration into simulation environments or driver assistance features.
Dataset Name Suggestions
- Traffic Sign Localization and Detection (YOLO Annotated)
- Annotated Road Sign Image Set (2.5K)
- YOLO-ready Traffic Signal Dataset
- Multiclass Road Sign Image Data
Attributes
Original Data Source: YOLO-ready Traffic Signal Dataset
QUALITYLoading...