Scholarlink

<p>Object detection in Unmanned Aerial Vehicle imagery is fundamentally challenged by a prevalence of small, densely packed, and occluded objects within cluttered backgrounds. Conventional detectors struggle with this domain, as they rely on hand-crafted components like pre-defined anchors and heuristic-based Non-Maximum Suppression, creating a well-known performance bottleneck in dense scenes. Even recent end-to-end frameworks have not been purpose-built to overcome these specific aerial challenges, resulting in a persistent performance gap. To bridge this gap, we introduce HEDS-DETR, a holistically enhanced real-time Detection Transformer tailored for aerial scenes. Our framework features three key innovations. First, we propose a novel High-Frequency Enhanced Semantics Network backbone, which yields highly discriminative features by preserving critical high-frequency details alongside robust semantic context. Second, our Efficient Small Object Pyramid counteracts information loss by efficiently fusing high-resolution features, significantly boosting small object detection. Finally, we enhance decoder stability and localization precision with two synergistic components: Selective Query Recollection and Geometry-Aware Positional Encoding, which stabilize optimization and provide explicit spatial priors for dense object arrangements. On the VisDrone dataset, HEDS-DETR achieves a <InlineEquation ID="IEq1"> <EquationSource Format="TEX">\(+3.8\%\)</EquationSource> </InlineEquation> AP and <InlineEquation ID="IEq2"> <EquationSource Format="TEX">\(+5.1\%\)</EquationSource> </InlineEquation> AP<InlineEquation ID="IEq3"> <EquationSource Format="TEX">\(_{50}\)</EquationSource> </InlineEquation> gain over its baseline while reducing parameters by 4&#xa0;M and maintaining real-time speeds. This demonstrates a highly competitive accuracy–efficiency balance, especially for detecting dense and small objects in aerial scenes.</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

High-frequency semantics and geometric priors for end-to-end detection transformers in challenging UAV imagery

  • 期刊论文

  • 发表日期：

    2026年5月15日

查看全文
  • Hongxing Peng,
  • Lide Chen,
  • Hui Zhu,
  • Yan Chen

摘要

Object detection in Unmanned Aerial Vehicle imagery is fundamentally challenged by a prevalence of small, densely packed, and occluded objects within cluttered backgrounds. Conventional detectors struggle with this domain, as they rely on hand-crafted components like pre-defined anchors and heuristic-based Non-Maximum Suppression, creating a well-known performance bottleneck in dense scenes. Even recent end-to-end frameworks have not been purpose-built to overcome these specific aerial challenges, resulting in a persistent performance gap. To bridge this gap, we introduce HEDS-DETR, a holistically enhanced real-time Detection Transformer tailored for aerial scenes. Our framework features three key innovations. First, we propose a novel High-Frequency Enhanced Semantics Network backbone, which yields highly discriminative features by preserving critical high-frequency details alongside robust semantic context. Second, our Efficient Small Object Pyramid counteracts information loss by efficiently fusing high-resolution features, significantly boosting small object detection. Finally, we enhance decoder stability and localization precision with two synergistic components: Selective Query Recollection and Geometry-Aware Positional Encoding, which stabilize optimization and provide explicit spatial priors for dense object arrangements. On the VisDrone dataset, HEDS-DETR achieves a \(+3.8\%\) AP and \(+5.1\%\) AP \(_{50}\) gain over its baseline while reducing parameters by 4 M and maintaining real-time speeds. This demonstrates a highly competitive accuracy–efficiency balance, especially for detecting dense and small objects in aerial scenes.