Scholarlink

<p>Small object detection in unmanned aerial vehicle (UAV) imagery is crucial for applications such as agricultural surveillance, infrastructure inspection, and traffic management. However, challenges like insufficient feature representation, background interference, and weak localization hinder detection accuracy. To address these issues, we propose CALENet, a context-aware and localization-enhanced detector. CALENet features a dual-path feature extractor (DPFE) for dynamic feature refinement, a perceptual enhancement module (PEM) for background suppression, and an enhanced localization fusion neck (ELFN) for accurate localization. Experimental results on the VisDrone2019 and AI-TODv2 datasets demonstrate significant performance improvements, with CALENet achieving an mAP50 increase of 8.8<InlineEquation ID="IEq1"> <EquationSource Format="TEX">\(\%\)</EquationSource> <EquationSource Format="MATHML"><math> <mo>%</mo> </math></EquationSource> </InlineEquation> and 5.3<InlineEquation ID="IEq2"> <EquationSource Format="TEX">\(\%\)</EquationSource> <EquationSource Format="MATHML"><math> <mo>%</mo> </math></EquationSource> </InlineEquation> improvement in mAP50-95 compared to the baseline model, while maintaining a low parameter count of 2.88 M. Our code is available: <a href="https://github.com/lqh964165950/CALENET">https://github.com/lqh964165950/CALENET</a>.</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 small object detection in aerial imagery through context-aware and localization-optimized deep learning

  • 期刊论文

  • 发表日期：

    2026年5月7日

查看全文
  • Qinghua Lai,
  • Lina Yang,
  • Haoyan Yang,
  • Xichun Li,
  • Qizong Lu,
  • Jiangtao Peng

摘要

Small object detection in unmanned aerial vehicle (UAV) imagery is crucial for applications such as agricultural surveillance, infrastructure inspection, and traffic management. However, challenges like insufficient feature representation, background interference, and weak localization hinder detection accuracy. To address these issues, we propose CALENet, a context-aware and localization-enhanced detector. CALENet features a dual-path feature extractor (DPFE) for dynamic feature refinement, a perceptual enhancement module (PEM) for background suppression, and an enhanced localization fusion neck (ELFN) for accurate localization. Experimental results on the VisDrone2019 and AI-TODv2 datasets demonstrate significant performance improvements, with CALENet achieving an mAP50 increase of 8.8 \(\%\) % and 5.3 \(\%\) % improvement in mAP50-95 compared to the baseline model, while maintaining a low parameter count of 2.88 M. Our code is available: https://github.com/lqh964165950/CALENET.