Scholarlink

<p>Recent advances in deep learning have significantly improved object detection performance. However, detecting small objects in drone-captured imagery remains challenging due to their low resolution and noisy appearance. This paper presents the Enhanced Single Shot Detector (ESSD), designed for accurate small object detection. The ESSD incorporates a scale-confusion erasing module to reduce noise from larger objects, enhancing the detection of smaller ones. It also features a neighbor fusion module that integrates semantic information across layers. Our experiments on the VisDrone2019-DET benchmark dataset show that the ESSD achieves state-of-the-art performance in small object detection.</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

Enhanced single shot detector for small object detection in drone-capture scenarios

  • 期刊论文

  • 发表日期：

    2025年2月1日

查看全文
  • Yanxia Shi,
  • Yanrong Liu,
  • Yaru Liu

摘要

Recent advances in deep learning have significantly improved object detection performance. However, detecting small objects in drone-captured imagery remains challenging due to their low resolution and noisy appearance. This paper presents the Enhanced Single Shot Detector (ESSD), designed for accurate small object detection. The ESSD incorporates a scale-confusion erasing module to reduce noise from larger objects, enhancing the detection of smaller ones. It also features a neighbor fusion module that integrates semantic information across layers. Our experiments on the VisDrone2019-DET benchmark dataset show that the ESSD achieves state-of-the-art performance in small object detection.