Augmenting Simulations for SAR ATR Neural Network Training

Authors

Spencer R. Sellers, Air Force Institute of Technology
Peter J. Collins, Air Force Institute of Technology
Julie Ann Jackson, Air Force Institute of TechnologyFollow

Document Type

Conference Proceeding

Publication Date

4-28-2020

Abstract

A training data augmentation technique is presented that approximates the differences between measured and simulated SAR imagery. This method is applied to simulated images and a CNN is trained with them. We achieve over 95% cross-class classification using the SAMPLE dataset from AFRL, with 1% measured data in the training set. We compare this to 89.6% accuracy when the augmentation technique is not used. Our hypothesis is that, while simulations can be made to approximate the measurements very closely, further augmentation can increase accuracy over non-augmented simulations.

Comments

The full text of this conference paper is available via subscription or purchase through the DOI link on this page.

DOI

10.1109/RADAR42522.2020.9114867

Source Publication

2020 IEEE International Radar Conference (RADAR)

Recommended Citation

S. R. Sellers, P. J. Collins and J. A. Jackson, "Augmenting Simulations for SAR ATR Neural Network Training," 2020 IEEE International Radar Conference (RADAR), Washington, DC, USA, 2020, pp. 309-314, doi: 10.1109/RADAR42522.2020.9114867