# AI-ML 开发者工作流

<?xml version="1.0" encoding="utf-8" standalone="no"?>
<!DOCTYPE svg PUBLIC "-//W3C//DTD SVG 1.1//EN" "http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd">
<!-- Generated by Microsoft Visio, SVG Export ai-developer-workflow.svg ai-ml-developer-workflow -->
<svg xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" xmlns:ev="http://www.w3.org/2001/xml-events" width="14.6082in" height="1.51042in" viewbox="0 0 1051.79 108.75" xml:space="preserve" color-interpolation-filters="sRGB" class="st6 svg-1" aria-label="../_images/ai-developer-workflow.svg" svgdefaultwidth="1191"><style>.svg-1 .st1 { fill: #2a2aea; stroke: #2a2aea; stroke-linecap: round; stroke-linejoin: round; stroke-width: 0.75 }
.svg-1 .st2 { fill: #ffffff; font-family: Roboto; font-size: 1.16666em; font-weight: bold }
.svg-1 .st3 { font-size: 1em }
.svg-1 .st4 { marker-end: url("#1-mrkr13-24"); stroke: #7030a0; stroke-linecap: round; stroke-linejoin: round; stroke-width: 1.5 }
.svg-1 .st5 { fill: #7030a0; fill-opacity: 1; stroke: #7030a0; stroke-opacity: 1; stroke-width: 0.37313432835821 }
.svg-1 .st6 { fill: none; fill-rule: evenodd; font-size: 12px; overflow: visible; stroke-linecap: square; stroke-miterlimit: 3 }</style>
<defs id="Markers">	<g id="lend13">		<path d="M 3 1 L 0 0 L 3 -1 L 3 1 " style="stroke:none"></path>	</g>	<marker id="1-mrkr13-24" class="st5" refx="-8.04" orient="auto" markerunits="strokeWidth" overflow="visible">		<use xlink:href="#lend13" transform="scale(-2.68,-2.68) "></use>	</marker></defs><g>	<title>ai-ml-developer-workflow</title>	<g id="shape27-1" transform="translate(18.375,-18.375)">		<title>Sheet.27</title>		<desc>Pretrained model</desc>		<rect x="0" y="36.75" width="151.038" height="72" rx="36" ry="36" class="st1"></rect>		<text x="21.15" y="76.95" class="st2">预训练模型</text>		</g>	<g id="shape28-4" transform="translate(234.375,-18.375)">		<title>Sheet.28</title>		<desc>Compile and optimize model</desc>		<rect x="0" y="36.75" width="151.038" height="72" rx="4.5" ry="4.5" class="st1"></rect>		<text x="36.12" y="68.55" class="st2">编译并优化模型<tspan x="25.05" dy="1.2em" class="st3"> </tspan></text>		</g>	<g id="shape29-8" transform="translate(450.375,-18.375)">		<title>Sheet.29</title>		<desc>Optimized model</desc>		<rect x="0" y="36.75" width="151.038" height="72" rx="36" ry="36" class="st1"></rect>		<text x="43.32" y="68.55" class="st2">优化后的<tspan x="55.92" dy="1.2em" class="st3">模型</tspan></text>		</g>	<g id="shape30-12" transform="translate(666.375,-18.375)">		<title>Sheet.30</title>		<desc>Build/customize AI/ML application</desc>		<rect x="0" y="36.75" width="151.038" height="72" rx="4.5" ry="4.5" class="st1"></rect>		<text x="24.11" y="68.55" class="st2">编译/定制<tspan x="19.64" dy="1.2em" class="st3">AI/ML 应用程序</tspan></text>		</g>	<g id="shape31-16" transform="translate(882.375,-18.375)">		<title>Sheet.31</title>		<desc>Deploy</desc>		<rect x="0" y="36.75" width="151.038" height="72" rx="36" ry="36" class="st1"></rect>		<text x="53.92" y="76.95" class="st2">部署</text>		</g>	<g id="shape32-19" transform="translate(169.386,-52.9811)">		<title>Sheet.32</title>		<path d="M0 108.75 L52.93 108.75" class="st4"></path>	</g>	<g id="shape33-25" transform="translate(385.413,-54)">		<title>Sheet.33</title>		<path d="M0 108.75 L53.01 108.75" class="st4"></path>	</g>	<g id="shape34-30" transform="translate(601.411,-54.0263)">		<title>Sheet.34</title>		<path d="M0 108.75 L52.9 108.75" class="st4"></path>	</g>	<g id="shape35-35" transform="translate(817.413,-54)">		<title>Sheet.35</title>		<path d="M0 108.75 L52.91 108.75" class="st4"></path>	</g></g>
</svg>

一起来观看本次培训课程，该培训重点介绍 Qualcomm Linux AI 开发者工作流。

它是专为物联网领域的开发者和专业人士量身定制的，满足对提高部署设备上 AI 解决方案的理解和技能。

<svg xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink" viewbox="0 0 640 400" width="640" height="400" style="cursor:auto !important" aria-label="../_images/ai-developer-workflow-video.svg" svgdefaultwidth="544">
    <defs>
      <style>@import url("https://fonts.googleapis.com/css2?family=Roboto+Flex:opsz,wght@8..144,100..1000&amp;display=swap");
.svg-2 .bg-fill { fill: var(--color-background) }
.svg-2 .fill-text { color: var(--color-content); fill: var(--color-content) }
.svg-2 .video-hoverbox { transition: opacity 0.15s ease-in-out }
.svg-2 .video-hoverbox:hover { opacity: 0.9 }</style>
  </defs>

  <foreignobject x="0" y="0" width="640" height="400">
    <body xmlns="http://www.w3.org/1999/xhtml">
        <iframe width="640" height="400" src="https://players.brightcove.net/1414329538001/4JiZQnWhg_default/index.html?videoId=6357317318112" allowfullscreen="" allow="encrypted-media"></iframe>
    <div class='topic-detail'><div class='topic-updated-date'><span> Last Published: </span>Jan 25, 2026</div><div class='prev-and-next-links'><span class='previous-topic-link'><span aria-hidden='true' class='disabled' data-tip='' data-effect='solid'></span></span></div></div></body>
    </foreignobject>
</svg>

Qualcomm Linux 中的 AI/ML 开发者工作流主要分为两个步骤：

| 第 1 步<br><br><br>编译并优化模型 | <ul class="simple"><br><li><p>编译并优化来自第三方 AI 框架的模型，以便在 Qualcomm 硬件上高效运行。例如，可以将 TensorFlow 模型导出为 TFLite 模型。</p></li><br><li><p>或者，使用硬件特定的定制对性能和精确度进行量化和微调。</p></li><br></ul> |
| --- | --- |
| 第 2 步<br><br><br>编译应用程序，使用优化后的模型运行设备推理 | <ul class="simple"><br><li><p>将 AI 模型集成到用例 pipeline 中。</p></li><br><li><p>交叉编译应用程序，用以生成使用依赖库的可执行二进制文件。</p></li><br></ul> |

Important

- 确保主机使用 Ubuntu 22.04。
- 本文档中的命令与 Qualcomm Linux 1.3 兼容。

    通过运行[开发套件快速入门指南](https://docs.qualcomm.com/bundle/publicresource/topics/80-70017-253/getting_started.html)中的命令来验证您的 Qualcomm Linux 发布版本。

    如果您的发布版本不是 1.3，请[更新软件](https://docs.qualcomm.com/bundle/publicresource/topics/80-70017-253/set_up_the_device.html#panel-0-VWJ1bnR1tab$update-software)。
- 本文档中的示例应用程序和 AI 程序与[支持的版本兼容](https://docs.qualcomm.com/bundle/publicresource/topics/80-70017-51/introduction.html#supported-component-versions)。

    在开始 AI/ML 开发之前，请确保将匹配的 SDK 下载到主机。

Last Published: Jan 25, 2026

[Next Topic
概述](https://docs.qualcomm.com/bundle/publicresource/80-70017-15BY/topics/overview.md)