Pothole Detection Data

Summary

• Introduction
• Dataset Structure
• Reference
• License

Introduction

This dataset is derived from multiple publicly available road inspection and traffic monitoring sources. It provides street-level images captured across diverse environments, including urban, rural, and highway road segments. The dataset covers varying lighting conditions such as daytime and nighttime, as well as different weather scenarios. Images are taken from multiple camera perspectives, including dashboard-mounted cameras, drones, and roadside CCTV.

The dataset is designed for the Pothole Detection Challenge, allowing participants to develop and evaluate computer vision models for automatic road damage detection.

Dataset Structure

Download

Download Pothole Detection Data

Include

The dataset will be released in two phases. Each phase will provide .zip files containing pothole image data with the following composition:

• Train set: Includes 145 pothole images, each with a corresponding .txt file containing label information about: "class x_center y_center box_width box_height":
  • class: object category (1 : pothole).
  • x_center, y_center: normalized coordinates of the bounding box center.
  • box_width, box_height: normalized width and height of the bounding box.
  Note: x_center and box_width are normalized by image_width; y_center and box_height are normalized by image_height.
• Test set: Includes 147 images, reserved for public scoring and evaluation.

Reference

For more information about the dataset, please visit the kaggle website.

License

This dataset is released under CC0: Public Domain