Data Augmentation via Mixed Class Interpolation using Cycle-Consistent Generative Adversarial Networks Applied to Cross-Domain Imagery

Sasaki, H.; Willcocks, C.G.; Breckon, T.P.

Abstract

Machine learning driven object detection and classification within non-visible imagery has an important role in many fields such as night vision, all-weather surveillance and aviation security. However, such applications often suffer due to the limited quantity and variety of non-visible spectral domain imagery, in contrast to the high data availability of visible-band imagery that readily enables contemporary deep learning driven detection and classification approaches. To address this problem, this paper proposes and evaluates a novel data augmentation approach that leverages the more readily available visible-band imagery via a generative domain transfer model. The model can synthesise large volumes of non-visible domain imagery by image-to-image (I2I) translation from the visible image domain. Furthermore, we show that the generation of interpolated mixed class (non-visible domain) image examples via our novel Conditional CycleGAN Mixup Augmentation (C2GMA) methodology can lead to a significant improvement in the quality of non-visible domain classification tasks that otherwise suffer due to limited data availability. Focusing on classification within the Synthetic Aperture Radar (SAR) domain, our approach is evaluated on a variation of the Statoil/C-CORE Iceberg Classifier Challenge dataset and achieves 75.4 % accuracy, demonstrating a significant improvement when compared against traditional data augmentation strategies (Rotation, Mixup, and MixCycleGAN).

Citation

Sasaki, H., Willcocks, C., & Breckon, T. (2021). Data Augmentation via Mixed Class Interpolation using Cycle-Consistent Generative Adversarial Networks Applied to Cross-Domain Imagery. . https://doi.org/10.1109/icpr48806.2021.9413023

• Conference Name: 25th International Conference on Pattern Recognition (ICPR 2020)
• Conference Location: Milan, Italy
• Start Date: Jan 10, 2021
• End Date: Jan 15, 2021
• Acceptance Date: Oct 11, 2020
• Online Publication Date: May 5, 2021
• Publication Date: 2021
• Deposit Date: Oct 25, 2020
• Publicly Available Date: Oct 27, 2020
• Series ISSN: 1051-4651
• DOI: https://doi.org/10.1109/icpr48806.2021.9413023