<rss xmlns:atom="http://www.w3.org/2005/Atom" version="2.0"><channel><title>OpenStreetMap - Tag - 堂堂一跑堂</title><link>https://spacetop.win/tags/openstreetmap/</link><description>OpenStreetMap - Tag - 堂堂一跑堂</description><generator>Hugo -- gohugo.io</generator><language>zh-CN</language><managingEditor>kingcopper@whu.edu.cn (WangTong)</managingEditor><webMaster>kingcopper@whu.edu.cn (WangTong)</webMaster><lastBuildDate>Sun, 31 May 2026 12:00:00 +0800</lastBuildDate><atom:link href="https://spacetop.win/tags/openstreetmap/" rel="self" type="application/rss+xml"/><item><title>GeoLink：用OpenStreetMap数据赋能遥感基础模型</title><link>https://spacetop.win/2026/05/20260531_153612_geolink_multimodal/</link><pubDate>Sun, 31 May 2026 12:00:00 +0800</pubDate><author><name>WangTong</name></author><guid>https://spacetop.win/2026/05/20260531_153612_geolink_multimodal/</guid><description><![CDATA[<h1 id="geolink用openstreetmap数据赋能遥感基础模型" class="headerLink">
    <a href="#geolink%e7%94%a8openstreetmap%e6%95%b0%e6%8d%ae%e8%b5%8b%e8%83%bd%e9%81%a5%e6%84%9f%e5%9f%ba%e7%a1%80%e6%a8%a1%e5%9e%8b" class="header-mark"></a>GeoLink：用OpenStreetMap数据赋能遥感基础模型</h1><blockquote>
  <p>📅 发表时间：2025年
🏛️ 会议：NeurIPS 2025
👥 作者：Lubian Bai, Xiuyuan Zhang, Siqi Zhang, Zepeng Zhang, Haoyu Wang, Wei Qin, Shihong Du
🔗 GitHub：https://github.com/bailubin/GeoLink_NeurIPS2025
📄 arXiv：https://arxiv.org/abs/2509.26016</p>
</blockquote><hr>
<h2 id="-论文信息" class="headerLink">
    <a href="#-%e8%ae%ba%e6%96%87%e4%bf%a1%e6%81%af" class="header-mark"></a>📌 论文信息</h2><p><strong>标题</strong>：GeoLink: Empowering Remote Sensing Foundation Model with OpenStreetMap Data</p>
<p><strong>关键词</strong>：遥感基础模型、OpenStreetMap、多模态融合、图神经网络、语义分割</p>
<p><strong>研究领域</strong>：遥感图像理解、地理空间人工智能、多模态学习</p>
<hr>
<h2 id="-问题背景遥感数据的单模态困境" class="headerLink">
    <a href="#-%e9%97%ae%e9%a2%98%e8%83%8c%e6%99%af%e9%81%a5%e6%84%9f%e6%95%b0%e6%8d%ae%e7%9a%84%e5%8d%95%e6%a8%a1%e6%80%81%e5%9b%b0%e5%a2%83" class="header-mark"></a>🔍 问题背景：遥感数据的&quot;单模态困境&quot;</h2><h3 id="核心问题" class="headerLink">
    <a href="#%e6%a0%b8%e5%bf%83%e9%97%ae%e9%a2%98" class="header-mark"></a>核心问题</h3><p>传统遥感基础模型存在一个根本性局限：<strong>只关注图像数据，忽略了其他地理空间数据源的互补信息</strong>。</p>
<h3 id="问题细节" class="headerLink">
    <a href="#%e9%97%ae%e9%a2%98%e7%bb%86%e8%8a%82" class="header-mark"></a>问题细节</h3><p>作者观察到一个关键现象：遥感图像和OpenStreetMap（OSM）数据提供了<strong>互补但异构</strong>的信息：</p>
<ol>
<li><strong>遥感图像</strong>：提供丰富的视觉特征（光谱、纹理、形状），但缺乏语义标注</li>
<li><strong>OSM数据</strong>：提供精确的语义信息（道路网络、建筑物轮廓、土地利用类型），但缺乏视觉细节</li>
</ol>
<h3 id="具体挑战" class="headerLink">
    <a href="#%e5%85%b7%e4%bd%93%e6%8c%91%e6%88%98" class="header-mark"></a>具体挑战</h3><p>作者从三个维度分析了这个&quot;模态鸿沟&quot;：</p>
<p><strong>数据结构异构性</strong>：</p>
<ul>
<li>遥感图像：规则的网格结构（pixel grid）</li>
<li>OSM数据：不规则的图结构（nodes, ways, relations）</li>
</ul>
<p><strong>语义粒度差异</strong>：</p>
<ul>
<li>遥感图像：像素级特征，需要后处理才能获得语义</li>
<li>OSM数据：对象级语义，直接带有类别标签</li>
</ul>
<p><strong>时空覆盖不一致</strong>：</p>
<ul>
<li>遥感图像：定期更新，但可能有云遮挡</li>
<li>OSM数据：众包更新，覆盖不均匀</li>
</ul>
<hr>
<h2 id="-解决方案geolink的三阶段融合框架" class="headerLink">
    <a href="#-%e8%a7%a3%e5%86%b3%e6%96%b9%e6%a1%88geolink%e7%9a%84%e4%b8%89%e9%98%b6%e6%ae%b5%e8%9e%8d%e5%90%88%e6%a1%86%e6%9e%b6" class="header-mark"></a>💡 解决方案：GeoLink的&quot;三阶段&quot;融合框架</h2><h3 id="核心思想" class="headerLink">
    <a href="#%e6%a0%b8%e5%bf%83%e6%80%9d%e6%83%b3" class="header-mark"></a>核心思想</h3><p>作者没有简单地将OSM数据作为额外输入通道，而是设计了一个<strong>层次化的多模态融合框架</strong>，在不同阶段整合两种模态的信息。</p>
<h3 id="技术细节" class="headerLink">
    <a href="#%e6%8a%80%e6%9c%af%e7%bb%86%e8%8a%82" class="header-mark"></a>技术细节</h3><h4 id="阶段1osm数据的图结构编码" class="headerLink">
    <a href="#%e9%98%b6%e6%ae%b51osm%e6%95%b0%e6%8d%ae%e7%9a%84%e5%9b%be%e7%bb%93%e6%9e%84%e7%bc%96%e7%a0%81" class="header-mark"></a>阶段1：OSM数据的图结构编码</h4><p><strong>关键创新</strong>：将OSM数据转换为异构图（Heterogeneous Graph）</p>
<div class="code-block highlight is-open show-line-numbers  tw-group tw-my-2">
  <div class="
    
    tw-flex 
    tw-flex-row
    tw-flex-1 
    tw-justify-between 
    tw-w-full tw-bg-bgColor-secondary
    ">      
    <button 
      class="
        code-block-button
        tw-mx-2 
        tw-flex
        tw-flex-row
        tw-flex-1"
      aria-hidden="true">
          <div class="group-[.is-open]:tw-rotate-90 tw-transition-[transform] tw-duration-500 tw-ease-in-out print:!tw-hidden tw-w-min tw-h-min tw-my-1 tw-mx-1"><svg class="icon"
    xmlns="http://www.w3.org/2000/svg" viewBox="0 0 320 512"><!-- Font Awesome Free 5.15.4 by @fontawesome - https://fontawesome.com License - https://fontawesome.com/license/free (Icons: CC BY 4.0, Fonts: SIL OFL 1.1, Code: MIT License) --><path d="M285.476 272.971L91.132 467.314c-9.373 9.373-24.569 9.373-33.941 0l-22.667-22.667c-9.357-9.357-9.375-24.522-.04-33.901L188.505 256 34.484 101.255c-9.335-9.379-9.317-24.544.04-33.901l22.667-22.667c9.373-9.373 24.569-9.373 33.941 0L285.475 239.03c9.373 9.372 9.373 24.568.001 33.941z"/></svg></div>
          <p class="tw-select-none !tw-my-1">text</p>]]></description></item><item><title>遥感基础模型新突破：SegEarth-OV与GeoLink的创新解读</title><link>https://spacetop.win/2026/05/20260531_140440_remote_sensing_foundation_model/</link><pubDate>Sun, 31 May 2026 12:00:00 +0800</pubDate><author><name>WangTong</name></author><guid>https://spacetop.win/2026/05/20260531_140440_remote_sensing_foundation_model/</guid><description><![CDATA[<h1 id="遥感基础模型新突破segearth-ov与geolink的创新解读" class="headerLink">
    <a href="#%e9%81%a5%e6%84%9f%e5%9f%ba%e7%a1%80%e6%a8%a1%e5%9e%8b%e6%96%b0%e7%aa%81%e7%a0%b4segearth-ov%e4%b8%8egeolink%e7%9a%84%e5%88%9b%e6%96%b0%e8%a7%a3%e8%af%bb" class="header-mark"></a>遥感基础模型新突破：SegEarth-OV与GeoLink的创新解读</h1><p><strong>关键词</strong>: 遥感基础模型, 开放词汇分割, 多模态融合, 无标注分割, OpenStreetMap, SAM3, CVPR 2025, NeurIPS 2025</p>
<hr>
<h2 id="一论文信息" class="headerLink">
    <a href="#%e4%b8%80%e8%ae%ba%e6%96%87%e4%bf%a1%e6%81%af" class="header-mark"></a>一、论文信息</h2><h3 id="论文1segearth-ov3---探索sam3在遥感开放词汇语义分割中的应用" class="headerLink">
    <a href="#%e8%ae%ba%e6%96%871segearth-ov3---%e6%8e%a2%e7%b4%a2sam3%e5%9c%a8%e9%81%a5%e6%84%9f%e5%bc%80%e6%94%be%e8%af%8d%e6%b1%87%e8%af%ad%e4%b9%89%e5%88%86%e5%89%b2%e4%b8%ad%e7%9a%84%e5%ba%94%e7%94%a8" class="header-mark"></a>论文1：SegEarth-OV3 - 探索SAM3在遥感开放词汇语义分割中的应用</h3><table>
  <thead>
      <tr>
          <th>项目</th>
          <th>内容</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td><strong>标题</strong></td>
          <td>SegEarth-OV3: Exploring SAM 3 for Open-Vocabulary Semantic Segmentation in Remote Sensing Images</td>
      </tr>
      <tr>
          <td><strong>作者</strong></td>
          <td>Kaiyu Li, Shengqi Zhang, Yupeng Deng, Zhi Wang, Deyu Meng, Xiangyong Cao</td>
      </tr>
      <tr>
          <td><strong>机构</strong></td>
          <td>西安交通大学, 中国科学院</td>
      </tr>
      <tr>
          <td><strong>发表</strong></td>
          <td>arXiv:2512.08730 (2025)</td>
      </tr>
      <tr>
          <td><strong>GitHub</strong></td>
          <td><a href="https://github.com/earth-insights/SegEarth-OV-3" target="_blank" rel="noopener noreferrer">https://github.com/earth-insights/SegEarth-OV-3</a> ⭐161</td>
      </tr>
      <tr>
          <td><strong>论文链接</strong></td>
          <td><a href="https://arxiv.org/abs/2512.08730" target="_blank" rel="noopener noreferrer">https://arxiv.org/abs/2512.08730</a></td>
      </tr>
  </tbody>
</table>
<h3 id="论文2geolink---利用openstreetmap数据增强遥感基础模型" class="headerLink">
    <a href="#%e8%ae%ba%e6%96%872geolink---%e5%88%a9%e7%94%a8openstreetmap%e6%95%b0%e6%8d%ae%e5%a2%9e%e5%bc%ba%e9%81%a5%e6%84%9f%e5%9f%ba%e7%a1%80%e6%a8%a1%e5%9e%8b" class="header-mark"></a>论文2：GeoLink - 利用OpenStreetMap数据增强遥感基础模型</h3><table>
  <thead>
      <tr>
          <th>项目</th>
          <th>内容</th>
      </tr>
  </thead>
  <tbody>
      <tr>
          <td><strong>标题</strong></td>
          <td>GeoLink: Empowering Remote Sensing Foundation Model with OpenStreetMap Data</td>
      </tr>
      <tr>
          <td><strong>作者</strong></td>
          <td>Lubian Bai, Xiuyuan Zhang, Siqi Zhang, Zepeng Zhang, Haoyu Wang, Wei Qin, Shihong Du</td>
      </tr>
      <tr>
          <td><strong>机构</strong></td>
          <td>北京大学</td>
      </tr>
      <tr>
          <td><strong>发表</strong></td>
          <td>NeurIPS 2025</td>
      </tr>
      <tr>
          <td><strong>GitHub</strong></td>
          <td><a href="https://github.com/bailubin/GeoLink_NeurIPS2025" target="_blank" rel="noopener noreferrer">https://github.com/bailubin/GeoLink_NeurIPS2025</a> ⭐56</td>
      </tr>
      <tr>
          <td><strong>论文链接</strong></td>
          <td><a href="https://arxiv.org/abs/2509.26016" target="_blank" rel="noopener noreferrer">https://arxiv.org/abs/2509.26016</a></td>
      </tr>
  </tbody>
</table>
<hr>
<h2 id="二问题背景与动机" class="headerLink">
    <a href="#%e4%ba%8c%e9%97%ae%e9%a2%98%e8%83%8c%e6%99%af%e4%b8%8e%e5%8a%a8%e6%9c%ba" class="header-mark"></a>二、问题背景与动机</h2><h3 id="21-遥感图像理解的核心挑战" class="headerLink">
    <a href="#21-%e9%81%a5%e6%84%9f%e5%9b%be%e5%83%8f%e7%90%86%e8%a7%a3%e7%9a%84%e6%a0%b8%e5%bf%83%e6%8c%91%e6%88%98" class="header-mark"></a>2.1 遥感图像理解的核心挑战</h3><p>遥感图像的语义分割是地球观测的关键任务，但面临两大根本性难题：</p>]]></description></item></channel></rss>