CrackAirport: A Dataset for Segmentation of Cracks in Airport Pavements

Description

CrackAirport contains 2,226 images of 512x512 pixels with annotations. These images include typical environmental patterns of airport pavements such as aircraft, T-hangars, vegetation, airport markings and signs, and marks of previous maintenance. The images were captured using a Sony ILCE-7RM4A camera mounted on a drone flying at an altitude of 100 feet. The original images were collected from various local airports in Tennessee. These images were annotated and then cropped to a size of 512x512 pixels. To balance the ratio of crack pixels to background pixels, some images with very few crack pixels were excluded. Additionally, some images with special noise were added to help a neural network, such as U-Net, recognize noise.

Institutions

• Tennessee Department of Transportation

  TN, Nashville
• Youngstown State University

  Ohio, Youngstown
• University of Tennessee at Knoxville

  Tennessee, Knoxville

Categories

Funders

• Tennessee Department of Transportation

  Tennessee, United States

Licence