A comparison of methods for generating synthetic training data for domain adaption of deep learning models in ultrasonic non-destructive evaluation

McKnight, Shaun and Pierce, S. Gareth and Mohseni, Ehsan and MacKinnon, Christopher and MacLeod, Charles and O'Hare, Tom and Loukas, Charalampos (2024) A comparison of methods for generating synthetic training data for domain adaption of deep learning models in ultrasonic non-destructive evaluation. NDT and E International, 141. 102978. ISSN 0963-8695 (https://doi.org/10.1016/j.ndteint.2023.102978)

[thumbnail of McKnight-etal-NDTEI-2023-A-comparison-of-methods-for-generating-synthetic-training-data-for-domain-adaption-of-deep-learning-models]
Preview
 Text. Filename: McKnight-etal-NDTEI-2023-A-comparison-of-methods-for-generating-synthetic-training-data-for-domain-adaption-of-deep-learning-models.pdf
Final Published Version
License: Creative Commons Attribution 4.0 logo
Download (10MB)| Preview

Abstract

This work provides a solution to the challenge of small amounts of training data in Non-Destructive Ultrasonic Testing for composite components. It was demonstrated that direct simulation alone is ineffective at producing training data that was representative of the experimental domain due to poor noise reconstruction. Therefore, four unique synthetic data generation methods were proposed which use semi-analytical simulated data as a foundation. Each method was evaluated for its performance in the classification of real experimental images when trained on a Convolutional Neural Network which underwent hyperparameter optimization using a genetic algorithm. The first method introduced task specific modifications to CycleGAN, a generative network for image-to-image translation, to learn the mapping from physics-based simulations of defect indications to experimental indications in resulting ultrasound images. The second method was based on combining real experimental defect free images with simulated defect responses. The final two methods fully simulated the noise responses at an image and signal level respectively. The purely simulated data produced a mean classification F1 score of 0.394. However, when trained on the new synthetic datasets, a significant improvement in classification performance on experimental data was realized, with mean classification F1 scores of 0.843, 0.688, 0.629, and 0.738 for the respective approaches.

CORE (COnnecting REpositories)