<rss xmlns:atom="http://www.w3.org/2005/Atom" version="2.0"><channel><title>时间一致性 - Tag - 堂堂一跑堂</title><link>https://spacetop.win/tags/%E6%97%B6%E9%97%B4%E4%B8%80%E8%87%B4%E6%80%A7/</link><description>时间一致性 - 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/%E6%97%B6%E9%97%B4%E4%B8%80%E8%87%B4%E6%80%A7/" rel="self" type="application/rss+xml"/><item><title>GeSCF：迈向可泛化的场景变化检测</title><link>https://spacetop.win/2026/05/20260531_120001_gescf/</link><pubDate>Sun, 31 May 2026 12:00:00 +0800</pubDate><author><name>WangTong</name></author><guid>https://spacetop.win/2026/05/20260531_120001_gescf/</guid><description><![CDATA[<h1 id="gescf迈向可泛化的场景变化检测" class="headerLink">
    <a href="#gescf%e8%bf%88%e5%90%91%e5%8f%af%e6%b3%9b%e5%8c%96%e7%9a%84%e5%9c%ba%e6%99%af%e5%8f%98%e5%8c%96%e6%a3%80%e6%b5%8b" class="header-mark"></a>GeSCF：迈向可泛化的场景变化检测</h1><blockquote>
  <p><strong>论文信息</strong></p>
<ul>
<li><strong>标题</strong>：Towards Generalizable Scene Change Detection</li>
<li><strong>作者</strong>：Jae-Woo Kim, Ue-Hwan Kim</li>
<li><strong>会议</strong>：CVPR 2025</li>
<li><strong>论文链接</strong>：https://arxiv.org/abs/2409.06214</li>
<li><strong>代码链接</strong>：https://github.com/AutoCompSysLab/towards-generalizable-scene-change-detection</li>
<li><strong>关键词</strong>：场景变化检测、零样本学习、Segment Anything Model、泛化性、时间一致性</li>
</ul>
</blockquote><hr>
<h2 id="一研究定位" class="headerLink">
    <a href="#%e4%b8%80%e7%a0%94%e7%a9%b6%e5%ae%9a%e4%bd%8d" class="header-mark"></a>一、研究定位</h2><h3 id="11-大领域" class="headerLink">
    <a href="#11-%e5%a4%a7%e9%a2%86%e5%9f%9f" class="header-mark"></a>1.1 大领域</h3><p>计算机视觉与遥感图像解译</p>
<h3 id="12-小领域" class="headerLink">
    <a href="#12-%e5%b0%8f%e9%a2%86%e5%9f%9f" class="header-mark"></a>1.2 小领域</h3><p>场景变化检测的泛化性问题——如何让变化检测模型在未见过的环境和时间条件下保持稳定性能</p>
<hr>
<h2 id="二研究问题从一个惊人的发现出发" class="headerLink">
    <a href="#%e4%ba%8c%e7%a0%94%e7%a9%b6%e9%97%ae%e9%a2%98%e4%bb%8e%e4%b8%80%e4%b8%aa%e6%83%8a%e4%ba%ba%e7%9a%84%e5%8f%91%e7%8e%b0%e5%87%ba%e5%8f%91" class="header-mark"></a>二、研究问题：从一个惊人的发现出发</h2><h3 id="21-问题来源" class="headerLink">
    <a href="#21-%e9%97%ae%e9%a2%98%e6%9d%a5%e6%ba%90" class="header-mark"></a>2.1 问题来源</h3><p>作者发现了一个被领域长期忽视的严重问题：<strong>现有场景变化检测（SCD）方法在研究数据上表现优异，但在真实世界中几乎失效</strong>。</p>
<p>具体而言，作者通过实验揭示了两个关键问题：</p>
<p><strong>问题一：域泛化性崩溃</strong></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></channel></rss>