Feature-Space Oversampling for Addressing Class Imbalance in SAR Ship Classification
This addresses the problem of long-tailed datasets for SAR ship classification, representing an incremental improvement with specific gains.
The paper tackled class imbalance in SAR ship classification by proposing two novel feature-space oversampling algorithms, M2m_f and M2m_u, which improved average F1-scores by 8.82% on FuSARShip and 4.44% on OpenSARShip datasets compared to baselines.
SAR ship classification faces the challenge of long-tailed datasets, which complicates the classification of underrepresented classes. Oversampling methods have proven effective in addressing class imbalance in optical data. In this paper, we evaluated the effect of oversampling in the feature space for SAR ship classification. We propose two novel algorithms inspired by the Major-to-minor (M2m) method M2m$_f$, M2m$_u$. The algorithms are tested on two public datasets, OpenSARShip (6 classes) and FuSARShip (9 classes), using three state-of-the-art models as feature extractors: ViT, VGG16, and ResNet50. Additionally, we also analyzed the impact of oversampling methods on different class sizes. The results demonstrated the effectiveness of our novel methods over the original M2m and baselines, with an average F1-score increase of 8.82% for FuSARShip and 4.44% for OpenSARShip.