Reliable research on hidden pipeline defects depends on more than innovative hardware. It also depends on data that can be examined, tested and reused. TUBERS has therefore made its Ultrasonic and Synthetic Crack Datasets openly available through the project’s Zenodo community.

Two complementary views of pipe condition

The release brings together two files: an ultrasonic dataset and a visual dataset of synthetic cracks. Together, they represent two complementary ways of looking at infrastructure condition. Visual data can capture surface-level signs of damage, while ultrasonic measurements can provide information about material condition that is not directly visible.

The complete record contains 46.3 MB of reusable research material. It was created by the University of Essex and TWI Hellas and is available under the MIT License.

Why open inspection data matters

Pipeline inspection research often faces a practical obstacle: high-quality defect data are difficult to obtain, especially for controlled comparisons. Real failures cannot be produced on demand, and utility data may be commercially sensitive or linked to critical infrastructure. Carefully generated experimental and synthetic data can help researchers develop and test methods before they are assessed in more varied real-world conditions.

Within TUBERS, visual and ultrasonic inputs form part of a wider effort to support automated defect detection and more informed maintenance decisions. Making these initial datasets public allows other researchers to examine the data, compare approaches and build on the project’s work.

A reusable foundation

This record is relevant to researchers working on non-destructive testing, computer vision, explainable machine learning and infrastructure monitoring. It also demonstrates TUBERS’ commitment to open science: project data should remain useful beyond the lifetime of the project itself.

Explore and download the TUBERS Ultrasonic and Synthetic Crack Datasets on Zenodo. Access the dataset on Zenodo