Scholarlink

<p>This study improves ship classification in Synthetic Aperture Radar (SAR) imagery, focusing on few-shot datasets. We propose a data augmentation strategy combining the AlignMixup method and a detail enhancement module to optimize convolutional neural network performance. AlignMixup integrates features at intermediate layers, capturing structural information, while the detail enhancement module highlights high-frequency details to improve ship feature recognition in SAR images. Experiments on small sample datasets show that our method increases classification accuracy by a significant margin and remains practical under data-limited conditions.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

AlignMixup-based ship classification in SAR imagery

  • 期刊论文

  • 发表日期：

    2025年1月28日

查看全文
  • Yongjin Chen,
  • Wentao An,
  • Bin Zou,
  • Peng Ren

摘要

This study improves ship classification in Synthetic Aperture Radar (SAR) imagery, focusing on few-shot datasets. We propose a data augmentation strategy combining the AlignMixup method and a detail enhancement module to optimize convolutional neural network performance. AlignMixup integrates features at intermediate layers, capturing structural information, while the detail enhancement module highlights high-frequency details to improve ship feature recognition in SAR images. Experiments on small sample datasets show that our method increases classification accuracy by a significant margin and remains practical under data-limited conditions.