Feature-Space Oversampling for Addressing Class Imbalance in SAR Ship Classification

2025-08-08

AI Summary

To address class imbalance caused by long-tailed distribution in SAR ship classification, this paper introduces the Major-to-minor (M2m) paradigm into feature-space oversampling for the first time, proposing two novel algorithms: M2m$_f$ (based on feature similarity) and M2m$_u$ (based on uncertainty-weighted synthesis). Extensive experiments are conducted using ViT, VGG16, and ResNet50 as backbone networks on OpenSARShip and FuSARShip datasets. Results demonstrate substantial improvements in minority-class recognition, with average F1-score gains of 4.44% and 8.82% on the respective datasets—outperforming both the original M2m method and state-of-the-art baselines. The core contribution lies in establishing a feature-space M2m oversampling framework that jointly preserves semantic plausibility and class discriminability, offering a lightweight, transferable solution for long-tailed SAR ship classification.

Technology Category

• Machine Learning: Multi-class/Multi-label Learning & Extreme Classification
• Computer Vision: Segmentation
• Multiagent Systems: Adversarial Agents

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: Federated recommendation systems and personalization
• Web Mining and Content Analysis: Normalization, clustering, classification, and summarization of Web text

Abstract

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.

Problem

• Addressing class imbalance in SAR ship classification
• Evaluating feature-space oversampling for underrepresented classes
• Proposing novel algorithms to improve F1-scores

Innovation

• Feature-space oversampling for SAR imbalance
• Novel M2m-inspired algorithms M2m_f, M2m_u
• Tested with ViT, VGG16, ResNet50 extractors