| |
| |
| |
| |
| |
| |
| <!DOCTYPE html> |
| <html class="writer-html5" lang="en" > |
| <head> |
| <meta charset="utf-8"> |
| |
| <meta name="viewport" content="width=device-width, initial-scale=1.0"> |
| |
| <title>Deploy Pretrained Vision Detection Model from Darknet on VTA — tvm 0.9.dev0 documentation</title> |
| |
| |
| |
| <link rel="stylesheet" href="https://maxcdn.bootstrapcdn.com/bootstrap/4.0.0/css/bootstrap.min.css" integrity="sha384-Gn5384xqQ1aoWXA+058RXPxPg6fy4IWvTNh0E263XmFcJlSAwiGgFAW/dAiS6JXm" crossorigin="anonymous"> |
| <link rel="stylesheet" href="../../../../_static/css/theme.css" type="text/css" /> |
| <link rel="stylesheet" href="../../../../_static/pygments.css" type="text/css" /> |
| <link rel="stylesheet" href="../../../../_static/css/theme.css" type="text/css" /> |
| <link rel="stylesheet" href="../../../../_static/gallery.css" type="text/css" /> |
| <link rel="stylesheet" href="../../../../_static/pygments.css" type="text/css" /> |
| <link rel="stylesheet" href="../../../../_static/css/tlcpack_theme.css" type="text/css" /> |
| |
| |
| |
| <link rel="shortcut icon" href="../../../../_static/tvm-logo-square.png"/> |
| |
| |
| |
| |
| |
| |
| |
| <script type="text/javascript" id="documentation_options" data-url_root="../../../../" src="../../../../_static/documentation_options.js"></script> |
| <script data-url_root="../../../../" id="documentation_options" src="../../../../_static/documentation_options.js"></script> |
| <script src="../../../../_static/jquery.js"></script> |
| <script src="../../../../_static/underscore.js"></script> |
| <script src="../../../../_static/doctools.js"></script> |
| |
| <script type="text/javascript" src="../../../../_static/js/theme.js"></script> |
| |
| |
| <script type="text/javascript" src="../../../../_static/js/tlcpack_theme.js"></script> |
| <link rel="index" title="Index" href="../../../../genindex.html" /> |
| <link rel="search" title="Search" href="../../../../search.html" /> |
| <link rel="next" title="2D Convolution Optimization" href="../optimize/convolution_opt.html" /> |
| <link rel="prev" title="Deploy Pretrained Vision Model from MxNet on VTA" href="deploy_classification.html" /> |
| </head> |
| |
| <body class="wy-body-for-nav"> |
| |
| |
| <div class="wy-grid-for-nav"> |
| |
| |
| <header class="header"> |
| <div class="innercontainer"> |
| <div class="headerInner d-flex justify-content-between align-items-center"> |
| <div class="headerLogo"> |
| <a href="https://tvm.apache.org/"><img src=https://tvm.apache.org/assets/images/logo.svg alt="logo"></a> |
| </div> |
| |
| <div id="headMenu" class="headerNav"> |
| <button type="button" id="closeHeadMenu" class="navCloseBtn"><img src="../../../../_static/img/close-icon.svg" alt="Close"></button> |
| <ul class="nav"> |
| <li class="nav-item"> |
| <a class="nav-link" href=https://tvm.apache.org/community>Community</a> |
| </li> |
| <li class="nav-item"> |
| <a class="nav-link" href=https://tvm.apache.org/download>Download</a> |
| </li> |
| <li class="nav-item"> |
| <a class="nav-link" href=https://tvm.apache.org/vta>VTA</a> |
| </li> |
| <li class="nav-item"> |
| <a class="nav-link" href=https://tvm.apache.org/blog>Blog</a> |
| </li> |
| <li class="nav-item"> |
| <a class="nav-link" href=https://tvm.apache.org/docs>Docs</a> |
| </li> |
| <li class="nav-item"> |
| <a class="nav-link" href=https://tvmconf.org>Conference</a> |
| </li> |
| <li class="nav-item"> |
| <a class="nav-link" href=https://github.com/apache/tvm/>Github</a> |
| </li> |
| </ul> |
| <div class="responsivetlcdropdown"> |
| <button type="button" class="btn-link"> |
| ASF |
| </button> |
| <ul> |
| <li> |
| <a href=https://apache.org/>Apache Homepage</a> |
| </li> |
| <li> |
| <a href=https://www.apache.org/licenses/>License</a> |
| </li> |
| <li> |
| <a href=https://www.apache.org/foundation/sponsorship.html>Sponsorship</a> |
| </li> |
| <li> |
| <a href=https://www.apache.org/security/>Security</a> |
| </li> |
| <li> |
| <a href=https://www.apache.org/foundation/thanks.html>Thanks</a> |
| </li> |
| <li> |
| <a href=https://www.apache.org/events/current-event>Events</a> |
| </li> |
| </ul> |
| </div> |
| </div> |
| <div class="responsiveMenuIcon"> |
| <button type="button" id="menuBtn" class="btn-menu"><img src="../../../../_static/img/menu-icon.svg" alt="Menu Icon"></button> |
| </div> |
| |
| <div class="tlcDropdown"> |
| <div class="dropdown"> |
| <button type="button" class="btn-link dropdown-toggle" data-toggle="dropdown" aria-haspopup="true" aria-expanded="false"> |
| ASF |
| </button> |
| <div class="dropdown-menu dropdown-menu-right"> |
| <ul> |
| <li> |
| <a href=https://apache.org/>Apache Homepage</a> |
| </li> |
| <li> |
| <a href=https://www.apache.org/licenses/>License</a> |
| </li> |
| <li> |
| <a href=https://www.apache.org/foundation/sponsorship.html>Sponsorship</a> |
| </li> |
| <li> |
| <a href=https://www.apache.org/security/>Security</a> |
| </li> |
| <li> |
| <a href=https://www.apache.org/foundation/thanks.html>Thanks</a> |
| </li> |
| <li> |
| <a href=https://www.apache.org/events/current-event>Events</a> |
| </li> |
| </ul> |
| </div> |
| </div> |
| </div> |
| </div> |
| </div> |
| </header> |
| |
| <nav data-toggle="wy-nav-shift" class="wy-nav-side fixed"> |
| <div class="wy-side-scroll"> |
| <div class="wy-side-nav-search" > |
| |
| |
| |
| <a href="../../../../index.html"> |
| |
| |
| |
| |
| <img src="../../../../_static/tvm-logo-small.png" class="logo" alt="Logo"/> |
| |
| </a> |
| |
| |
| |
| |
| <div class="version"> |
| 0.9.dev0 |
| </div> |
| |
| |
| |
| |
| <div role="search"> |
| <form id="rtd-search-form" class="wy-form" action="../../../../search.html" method="get"> |
| <input type="text" name="q" placeholder="Search docs" /> |
| <input type="hidden" name="check_keywords" value="yes" /> |
| <input type="hidden" name="area" value="default" /> |
| </form> |
| </div> |
| |
| |
| </div> |
| |
| |
| <div class="wy-menu wy-menu-vertical" data-spy="affix" role="navigation" aria-label="main navigation"> |
| |
| |
| |
| |
| |
| |
| <p class="caption" role="heading"><span class="caption-text">Getting Started</span></p> |
| <ul> |
| <li class="toctree-l1"><a class="reference internal" href="../../../../install/index.html">Installing TVM</a></li> |
| <li class="toctree-l1"><a class="reference internal" href="../../../../contribute/index.html">Contributor Guide</a></li> |
| </ul> |
| <p class="caption" role="heading"><span class="caption-text">User Guide</span></p> |
| <ul> |
| <li class="toctree-l1"><a class="reference internal" href="../../../../tutorial/index.html">User Tutorial</a></li> |
| <li class="toctree-l1"><a class="reference internal" href="../../../../how_to/index.html">How To Guides</a></li> |
| </ul> |
| <p class="caption" role="heading"><span class="caption-text">Developer Guide</span></p> |
| <ul> |
| <li class="toctree-l1"><a class="reference internal" href="../../../../dev/tutorial/index.html">Developer Tutorial</a></li> |
| <li class="toctree-l1"><a class="reference internal" href="../../../../dev/how_to/how_to.html">Developer How-To Guide</a></li> |
| </ul> |
| <p class="caption" role="heading"><span class="caption-text">Architecture Guide</span></p> |
| <ul> |
| <li class="toctree-l1"><a class="reference internal" href="../../../../arch/index.html">Design and Architecture</a></li> |
| </ul> |
| <p class="caption" role="heading"><span class="caption-text">Topic Guides</span></p> |
| <ul class="current"> |
| <li class="toctree-l1"><a class="reference internal" href="../../../microtvm/index.html">microTVM: TVM on bare-metal</a></li> |
| <li class="toctree-l1 current"><a class="reference internal" href="../../index.html">VTA: Versatile Tensor Accelerator</a><ul class="current"> |
| <li class="toctree-l2"><a class="reference internal" href="../../install.html">VTA Installation Guide</a></li> |
| <li class="toctree-l2"><a class="reference internal" href="../../dev/index.html">VTA Design and Developer Guide</a></li> |
| <li class="toctree-l2 current"><a class="reference internal" href="../index.html">VTA Tutorials</a><ul class="current"> |
| <li class="toctree-l3"><a class="reference internal" href="../matrix_multiply.html">Simple Matrix Multiply</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../vta_get_started.html">Get Started with VTA</a></li> |
| <li class="toctree-l3 current"><a class="reference internal" href="../index.html#compile-deep-learning-models">Compile Deep Learning Models</a><ul class="current"> |
| <li class="toctree-l4"><a class="reference internal" href="deploy_classification.html">Deploy Pretrained Vision Model from MxNet on VTA</a></li> |
| <li class="toctree-l4 current"><a class="current reference internal" href="#">Deploy Pretrained Vision Detection Model from Darknet on VTA</a></li> |
| </ul> |
| </li> |
| <li class="toctree-l3"><a class="reference internal" href="../index.html#optimize-tensor-operators">Optimize Tensor Operators</a></li> |
| <li class="toctree-l3"><a class="reference internal" href="../index.html#auto-tuning">Auto tuning</a></li> |
| </ul> |
| </li> |
| <li class="toctree-l2"><a class="reference internal" href="../../index.html#literature">Literature</a></li> |
| </ul> |
| </li> |
| </ul> |
| <p class="caption" role="heading"><span class="caption-text">Reference Guide</span></p> |
| <ul> |
| <li class="toctree-l1"><a class="reference internal" href="../../../../reference/langref/index.html">Language Reference</a></li> |
| <li class="toctree-l1"><a class="reference internal" href="../../../../reference/api/python/index.html">Python API</a></li> |
| <li class="toctree-l1"><a class="reference internal" href="../../../../reference/api/links.html">Other APIs</a></li> |
| <li class="toctree-l1"><a class="reference internal" href="../../../../reference/publications.html">Publications</a></li> |
| <li class="toctree-l1"><a class="reference internal" href="../../../../genindex.html">Index</a></li> |
| </ul> |
| |
| |
| |
| </div> |
| |
| </div> |
| </nav> |
| |
| <section data-toggle="wy-nav-shift" class="wy-nav-content-wrap"> |
| |
| <nav class="wy-nav-top" aria-label="top navigation" data-toggle="wy-nav-top"> |
| |
| <div class="togglemenu"> |
| |
| </div> |
| <div class="nav-content"> |
| <!-- tvm --> |
| Table of content |
| </div> |
| |
| </nav> |
| |
| |
| <div class="wy-nav-content"> |
| |
| <div class="rst-content"> |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| <div role="navigation" aria-label="breadcrumbs navigation"> |
| |
| <ul class="wy-breadcrumbs"> |
| |
| <li><a href="../../../../index.html">Docs</a> <span class="br-arrow">></span></li> |
| |
| <li><a href="../../index.html">VTA: Versatile Tensor Accelerator</a> <span class="br-arrow">></span></li> |
| |
| <li><a href="../index.html">VTA Tutorials</a> <span class="br-arrow">></span></li> |
| |
| <li>Deploy Pretrained Vision Detection Model from Darknet on VTA</li> |
| |
| |
| <li class="wy-breadcrumbs-aside"> |
| |
| |
| <a href="../../../../_sources/topic/vta/tutorials/frontend/deploy_detection.rst.txt" rel="nofollow"> <img src="../../../../_static//img/source.svg" alt="viewsource"/></a> |
| |
| |
| </li> |
| |
| </ul> |
| |
| |
| <hr/> |
| </div> |
| <div role="main" class="document" itemscope="itemscope" itemtype="http://schema.org/Article"> |
| <div itemprop="articleBody"> |
| |
| <div class="sphx-glr-download-link-note admonition note"> |
| <p class="admonition-title">Note</p> |
| <p>Click <a class="reference internal" href="#sphx-glr-download-topic-vta-tutorials-frontend-deploy-detection-py"><span class="std std-ref">here</span></a> to download the full example code</p> |
| </div> |
| <div class="sphx-glr-example-title section" id="deploy-pretrained-vision-detection-model-from-darknet-on-vta"> |
| <span id="sphx-glr-topic-vta-tutorials-frontend-deploy-detection-py"></span><h1>Deploy Pretrained Vision Detection Model from Darknet on VTA<a class="headerlink" href="#deploy-pretrained-vision-detection-model-from-darknet-on-vta" title="Permalink to this headline">¶</a></h1> |
| <p><strong>Author</strong>: <a class="reference external" href="https://github.com/huajsj">Hua Jiang</a></p> |
| <p>This tutorial provides an end-to-end demo, on how to run Darknet YoloV3-tiny |
| inference onto the VTA accelerator design to perform Image detection tasks. |
| It showcases Relay as a front end compiler that can perform quantization (VTA |
| only supports int8/32 inference) as well as graph packing (in order to enable |
| tensorization in the core) to massage the compute graph for the hardware target.</p> |
| <div class="section" id="install-dependencies"> |
| <h2>Install dependencies<a class="headerlink" href="#install-dependencies" title="Permalink to this headline">¶</a></h2> |
| <p>To use the autotvm package in tvm, we need to install some extra dependencies. |
| (change “3” to “2” if you use python2):</p> |
| <div class="highlight-bash notranslate"><div class="highlight"><pre><span></span>pip3 install <span class="s2">"Pillow<7"</span> |
| </pre></div> |
| </div> |
| <p>YOLO-V3-tiny Model with Darknet parsing have dependancy with CFFI and CV2 library, |
| we need to install CFFI and CV2 before executing this script.</p> |
| <div class="highlight-bash notranslate"><div class="highlight"><pre><span></span>pip3 install cffi |
| pip3 install opencv-python |
| </pre></div> |
| </div> |
| <p>Now return to the python code. Import packages.</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="kn">from</span> <span class="nn">__future__</span> <span class="k">import</span> <span class="n">absolute_import</span><span class="p">,</span> <span class="n">print_function</span> |
| |
| <span class="kn">import</span> <span class="nn">sys</span> |
| <span class="kn">import</span> <span class="nn">os</span> |
| <span class="kn">import</span> <span class="nn">time</span> |
| <span class="kn">import</span> <span class="nn">matplotlib.pyplot</span> <span class="k">as</span> <span class="nn">plt</span> |
| <span class="kn">import</span> <span class="nn">numpy</span> <span class="k">as</span> <span class="nn">np</span> |
| <span class="kn">import</span> <span class="nn">tvm</span> |
| <span class="kn">import</span> <span class="nn">vta</span> |
| <span class="kn">from</span> <span class="nn">tvm</span> <span class="k">import</span> <span class="n">rpc</span><span class="p">,</span> <span class="n">autotvm</span><span class="p">,</span> <span class="n">relay</span> |
| <span class="kn">from</span> <span class="nn">tvm.relay.testing</span> <span class="k">import</span> <span class="n">yolo_detection</span><span class="p">,</span> <span class="n">darknet</span> |
| <span class="kn">from</span> <span class="nn">tvm.relay.testing.darknet</span> <span class="k">import</span> <span class="n">__darknetffi__</span> |
| <span class="kn">from</span> <span class="nn">tvm.contrib</span> <span class="k">import</span> <span class="n">graph_executor</span><span class="p">,</span> <span class="n">utils</span> |
| <span class="kn">from</span> <span class="nn">tvm.contrib.download</span> <span class="k">import</span> <span class="n">download_testdata</span> |
| <span class="kn">from</span> <span class="nn">vta.testing</span> <span class="k">import</span> <span class="n">simulator</span> |
| <span class="kn">from</span> <span class="nn">vta.top</span> <span class="k">import</span> <span class="n">graph_pack</span> |
| |
| <span class="c1"># Make sure that TVM was compiled with RPC=1</span> |
| <span class="k">assert</span> <a href="../../../../reference/api/python/runtime.html#tvm.runtime.enabled" title="View documentation for tvm.runtime.enabled"><span class="n">tvm</span><span class="o">.</span><span class="n">runtime</span><span class="o">.</span><span class="n">enabled</span></a><span class="p">(</span><span class="s2">"rpc"</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| <p>Download yolo net configure file, weight file, darknet library file based on |
| Model Name |
| —————————————————————————-</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">MODEL_NAME</span> <span class="o">=</span> <span class="s2">"yolov3-tiny"</span> |
| <span class="n">REPO_URL</span> <span class="o">=</span> <span class="s2">"https://github.com/dmlc/web-data/blob/main/darknet/"</span> |
| |
| <span class="n">cfg_path</span> <span class="o">=</span> <span class="n">download_testdata</span><span class="p">(</span> |
| <span class="s2">"https://github.com/pjreddie/darknet/blob/master/cfg/"</span> <span class="o">+</span> <span class="n">MODEL_NAME</span> <span class="o">+</span> <span class="s2">".cfg"</span> <span class="o">+</span> <span class="s2">"?raw=true"</span><span class="p">,</span> |
| <span class="n">MODEL_NAME</span> <span class="o">+</span> <span class="s2">".cfg"</span><span class="p">,</span> |
| <span class="n">module</span><span class="o">=</span><span class="s2">"darknet"</span><span class="p">,</span> |
| <span class="p">)</span> |
| <span class="n">weights_path</span> <span class="o">=</span> <span class="n">download_testdata</span><span class="p">(</span> |
| <span class="s2">"https://pjreddie.com/media/files/"</span> <span class="o">+</span> <span class="n">MODEL_NAME</span> <span class="o">+</span> <span class="s2">".weights"</span> <span class="o">+</span> <span class="s2">"?raw=true"</span><span class="p">,</span> |
| <span class="n">MODEL_NAME</span> <span class="o">+</span> <span class="s2">".weights"</span><span class="p">,</span> |
| <span class="n">module</span><span class="o">=</span><span class="s2">"darknet"</span><span class="p">,</span> |
| <span class="p">)</span> |
| |
| <span class="k">if</span> <span class="n">sys</span><span class="o">.</span><span class="n">platform</span> <span class="ow">in</span> <span class="p">[</span><span class="s2">"linux"</span><span class="p">,</span> <span class="s2">"linux2"</span><span class="p">]:</span> |
| <span class="n">darknet_lib_path</span> <span class="o">=</span> <span class="n">download_testdata</span><span class="p">(</span> |
| <span class="n">REPO_URL</span> <span class="o">+</span> <span class="s2">"lib/"</span> <span class="o">+</span> <span class="s2">"libdarknet2.0.so"</span> <span class="o">+</span> <span class="s2">"?raw=true"</span><span class="p">,</span> <span class="s2">"libdarknet2.0.so"</span><span class="p">,</span> <span class="n">module</span><span class="o">=</span><span class="s2">"darknet"</span> |
| <span class="p">)</span> |
| <span class="k">elif</span> <span class="n">sys</span><span class="o">.</span><span class="n">platform</span> <span class="o">==</span> <span class="s2">"darwin"</span><span class="p">:</span> |
| <span class="n">darknet_lib_path</span> <span class="o">=</span> <span class="n">download_testdata</span><span class="p">(</span> |
| <span class="n">REPO_URL</span> <span class="o">+</span> <span class="s2">"lib_osx/"</span> <span class="o">+</span> <span class="s2">"libdarknet_mac2.0.so"</span> <span class="o">+</span> <span class="s2">"?raw=true"</span><span class="p">,</span> |
| <span class="s2">"libdarknet_mac2.0.so"</span><span class="p">,</span> |
| <span class="n">module</span><span class="o">=</span><span class="s2">"darknet"</span><span class="p">,</span> |
| <span class="p">)</span> |
| <span class="k">else</span><span class="p">:</span> |
| <span class="k">raise</span> <span class="ne">NotImplementedError</span><span class="p">(</span><span class="s2">"Darknet lib is not supported on </span><span class="si">{}</span><span class="s2"> platform"</span><span class="o">.</span><span class="n">format</span><span class="p">(</span><span class="n">sys</span><span class="o">.</span><span class="n">platform</span><span class="p">))</span> |
| </pre></div> |
| </div> |
| </div> |
| <div class="section" id="download-yolo-categories-and-illustration-front"> |
| <h2>Download yolo categories and illustration front.<a class="headerlink" href="#download-yolo-categories-and-illustration-front" title="Permalink to this headline">¶</a></h2> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="n">coco_path</span> <span class="o">=</span> <span class="n">download_testdata</span><span class="p">(</span> |
| <span class="n">REPO_URL</span> <span class="o">+</span> <span class="s2">"data/"</span> <span class="o">+</span> <span class="s2">"coco.names"</span> <span class="o">+</span> <span class="s2">"?raw=true"</span><span class="p">,</span> <span class="s2">"coco.names"</span><span class="p">,</span> <span class="n">module</span><span class="o">=</span><span class="s2">"data"</span> |
| <span class="p">)</span> |
| <span class="n">font_path</span> <span class="o">=</span> <span class="n">download_testdata</span><span class="p">(</span> |
| <span class="n">REPO_URL</span> <span class="o">+</span> <span class="s2">"data/"</span> <span class="o">+</span> <span class="s2">"arial.ttf"</span> <span class="o">+</span> <span class="s2">"?raw=true"</span><span class="p">,</span> <span class="s2">"arial.ttf"</span><span class="p">,</span> <span class="n">module</span><span class="o">=</span><span class="s2">"data"</span> |
| <span class="p">)</span> |
| <span class="k">with</span> <span class="nb">open</span><span class="p">(</span><span class="n">coco_path</span><span class="p">)</span> <span class="k">as</span> <span class="n">f</span><span class="p">:</span> |
| <span class="n">content</span> <span class="o">=</span> <span class="n">f</span><span class="o">.</span><span class="n">readlines</span><span class="p">()</span> |
| <span class="n">names</span> <span class="o">=</span> <span class="p">[</span><span class="n">x</span><span class="o">.</span><span class="n">strip</span><span class="p">()</span> <span class="k">for</span> <span class="n">x</span> <span class="ow">in</span> <span class="n">content</span><span class="p">]</span> |
| </pre></div> |
| </div> |
| </div> |
| <div class="section" id="define-the-platform-and-model-targets"> |
| <h2>Define the platform and model targets.<a class="headerlink" href="#define-the-platform-and-model-targets" title="Permalink to this headline">¶</a></h2> |
| <p>Execute on CPU vs. VTA, and define the model.</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="c1"># Load VTA parameters from the 3rdparty/vta-hw/config/vta_config.json file</span> |
| <span class="n">env</span> <span class="o">=</span> <span class="n">vta</span><span class="o">.</span><span class="n">get_env</span><span class="p">()</span> |
| <span class="c1"># Set ``device=arm_cpu`` to run inference on the CPU</span> |
| <span class="c1"># or ``device=vta`` to run inference on the FPGA.</span> |
| <span class="n">device</span> <span class="o">=</span> <span class="s2">"vta"</span> |
| <span class="n">target</span> <span class="o">=</span> <span class="n">env</span><span class="o">.</span><span class="n">target</span> <span class="k">if</span> <span class="n">device</span> <span class="o">==</span> <span class="s2">"vta"</span> <span class="k">else</span> <span class="n">env</span><span class="o">.</span><span class="n">target_vta_cpu</span> |
| |
| <span class="n">pack_dict</span> <span class="o">=</span> <span class="p">{</span> |
| <span class="s2">"yolov3-tiny"</span><span class="p">:</span> <span class="p">[</span><span class="s2">"nn.max_pool2d"</span><span class="p">,</span> <span class="s2">"cast"</span><span class="p">,</span> <span class="mi">4</span><span class="p">,</span> <span class="mi">186</span><span class="p">],</span> |
| <span class="p">}</span> |
| |
| <span class="c1"># Name of Darknet model to compile</span> |
| <span class="c1"># The ``start_pack`` and ``stop_pack`` labels indicate where</span> |
| <span class="c1"># to start and end the graph packing relay pass: in other words</span> |
| <span class="c1"># where to start and finish offloading to VTA.</span> |
| <span class="c1"># the number 4 indicate the the ``start_pack`` index is 4, the</span> |
| <span class="c1"># number 186 indicate the ``stop_pack index`` is 186, by using</span> |
| <span class="c1"># name and index number, here we can located to correct place</span> |
| <span class="c1"># where to start/end when there are multiple ``nn.max_pool2d``</span> |
| <span class="c1"># or ``cast``, print(mod.astext(show_meta_data=False)) can help</span> |
| <span class="c1"># to find operator name and index information.</span> |
| <span class="k">assert</span> <span class="n">MODEL_NAME</span> <span class="ow">in</span> <span class="n">pack_dict</span> |
| </pre></div> |
| </div> |
| </div> |
| <div class="section" id="obtain-an-execution-remote"> |
| <h2>Obtain an execution remote.<a class="headerlink" href="#obtain-an-execution-remote" title="Permalink to this headline">¶</a></h2> |
| <p>When target is ‘pynq’ or other FPGA backend, reconfigure FPGA and runtime. |
| Otherwise, if target is ‘sim’, execute locally.</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="k">if</span> <span class="n">env</span><span class="o">.</span><span class="n">TARGET</span> <span class="ow">not</span> <span class="ow">in</span> <span class="p">[</span><span class="s2">"sim"</span><span class="p">,</span> <span class="s2">"tsim"</span><span class="p">]:</span> |
| <span class="c1"># Get remote from tracker node if environment variable is set.</span> |
| <span class="c1"># To set up the tracker, you'll need to follow the "Auto-tuning</span> |
| <span class="c1"># a convolutional network for VTA" tutorial.</span> |
| <span class="n">tracker_host</span> <span class="o">=</span> <span class="n">os</span><span class="o">.</span><span class="n">environ</span><span class="o">.</span><span class="n">get</span><span class="p">(</span><span class="s2">"TVM_TRACKER_HOST"</span><span class="p">,</span> <span class="kc">None</span><span class="p">)</span> |
| <span class="n">tracker_port</span> <span class="o">=</span> <span class="n">os</span><span class="o">.</span><span class="n">environ</span><span class="o">.</span><span class="n">get</span><span class="p">(</span><span class="s2">"TVM_TRACKER_PORT"</span><span class="p">,</span> <span class="kc">None</span><span class="p">)</span> |
| <span class="c1"># Otherwise if you have a device you want to program directly from</span> |
| <span class="c1"># the host, make sure you've set the variables below to the IP of</span> |
| <span class="c1"># your board.</span> |
| <span class="n">device_host</span> <span class="o">=</span> <span class="n">os</span><span class="o">.</span><span class="n">environ</span><span class="o">.</span><span class="n">get</span><span class="p">(</span><span class="s2">"VTA_RPC_HOST"</span><span class="p">,</span> <span class="s2">"192.168.2.99"</span><span class="p">)</span> |
| <span class="n">device_port</span> <span class="o">=</span> <span class="n">os</span><span class="o">.</span><span class="n">environ</span><span class="o">.</span><span class="n">get</span><span class="p">(</span><span class="s2">"VTA_RPC_PORT"</span><span class="p">,</span> <span class="s2">"9091"</span><span class="p">)</span> |
| <span class="k">if</span> <span class="ow">not</span> <span class="n">tracker_host</span> <span class="ow">or</span> <span class="ow">not</span> <span class="n">tracker_port</span><span class="p">:</span> |
| <span class="n">remote</span> <span class="o">=</span> <a href="../../../../reference/api/python/rpc.html#tvm.rpc.connect" title="View documentation for tvm.rpc.connect"><span class="n">rpc</span><span class="o">.</span><span class="n">connect</span></a><span class="p">(</span><span class="n">device_host</span><span class="p">,</span> <span class="nb">int</span><span class="p">(</span><span class="n">device_port</span><span class="p">))</span> |
| <span class="k">else</span><span class="p">:</span> |
| <span class="n">remote</span> <span class="o">=</span> <span class="n">autotvm</span><span class="o">.</span><span class="n">measure</span><span class="o">.</span><span class="n">request_remote</span><span class="p">(</span> |
| <span class="n">env</span><span class="o">.</span><span class="n">TARGET</span><span class="p">,</span> <span class="n">tracker_host</span><span class="p">,</span> <span class="nb">int</span><span class="p">(</span><span class="n">tracker_port</span><span class="p">),</span> <span class="n">timeout</span><span class="o">=</span><span class="mi">10000</span> |
| <span class="p">)</span> |
| <span class="c1"># Reconfigure the JIT runtime and FPGA.</span> |
| <span class="c1"># You can program the FPGA with your own custom bitstream</span> |
| <span class="c1"># by passing the path to the bitstream file instead of None.</span> |
| <span class="n">reconfig_start</span> <span class="o">=</span> <span class="n">time</span><span class="o">.</span><span class="n">time</span><span class="p">()</span> |
| <span class="n">vta</span><span class="o">.</span><span class="n">reconfig_runtime</span><span class="p">(</span><span class="n">remote</span><span class="p">)</span> |
| <span class="n">vta</span><span class="o">.</span><span class="n">program_fpga</span><span class="p">(</span><span class="n">remote</span><span class="p">,</span> <span class="n">bitstream</span><span class="o">=</span><span class="kc">None</span><span class="p">)</span> |
| <span class="n">reconfig_time</span> <span class="o">=</span> <span class="n">time</span><span class="o">.</span><span class="n">time</span><span class="p">()</span> <span class="o">-</span> <span class="n">reconfig_start</span> |
| <span class="nb">print</span><span class="p">(</span><span class="s2">"Reconfigured FPGA and RPC runtime in </span><span class="si">{0:.2f}</span><span class="s2">s!"</span><span class="o">.</span><span class="n">format</span><span class="p">(</span><span class="n">reconfig_time</span><span class="p">))</span> |
| |
| <span class="c1"># In simulation mode, host the RPC server locally.</span> |
| <span class="k">else</span><span class="p">:</span> |
| <span class="n">remote</span> <span class="o">=</span> <a href="../../../../reference/api/python/rpc.html#tvm.rpc.LocalSession" title="View documentation for tvm.rpc.LocalSession"><span class="n">rpc</span><span class="o">.</span><span class="n">LocalSession</span></a><span class="p">()</span> |
| |
| <span class="c1"># Get execution context from remote</span> |
| <span class="n">ctx</span> <span class="o">=</span> <span class="n">remote</span><span class="o">.</span><span class="n">ext_dev</span><span class="p">(</span><span class="mi">0</span><span class="p">)</span> <span class="k">if</span> <span class="n">device</span> <span class="o">==</span> <span class="s2">"vta"</span> <span class="k">else</span> <span class="n">remote</span><span class="o">.</span><span class="n">cpu</span><span class="p">(</span><span class="mi">0</span><span class="p">)</span> |
| </pre></div> |
| </div> |
| </div> |
| <div class="section" id="build-the-inference-graph-executor"> |
| <h2>Build the inference graph executor.<a class="headerlink" href="#build-the-inference-graph-executor" title="Permalink to this headline">¶</a></h2> |
| <p>Using Darknet library load downloaded vision model and compile with Relay. |
| The compilation steps are:</p> |
| <ol class="arabic simple"> |
| <li><p>Front end translation from Darknet into Relay module.</p></li> |
| <li><p>Apply 8-bit quantization: here we skip the first conv layer, |
| and dense layer which will both be executed in fp32 on the CPU.</p></li> |
| <li><p>Perform graph packing to alter the data layout for tensorization.</p></li> |
| <li><p>Perform constant folding to reduce number of operators (e.g. eliminate batch norm multiply).</p></li> |
| <li><p>Perform relay build to object file.</p></li> |
| <li><p>Load the object file onto remote (FPGA device).</p></li> |
| <li><p>Generate graph executor, <cite>m</cite>.</p></li> |
| </ol> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="c1"># Load pre-configured AutoTVM schedules</span> |
| <span class="k">with</span> <span class="n">autotvm</span><span class="o">.</span><span class="n">tophub</span><span class="o">.</span><span class="n">context</span><span class="p">(</span><span class="n">target</span><span class="p">):</span> |
| <span class="n">net</span> <span class="o">=</span> <span class="n">__darknetffi__</span><span class="o">.</span><span class="n">dlopen</span><span class="p">(</span><span class="n">darknet_lib_path</span><span class="p">)</span><span class="o">.</span><span class="n">load_network</span><span class="p">(</span> |
| <span class="n">cfg_path</span><span class="o">.</span><span class="n">encode</span><span class="p">(</span><span class="s2">"utf-8"</span><span class="p">),</span> <span class="n">weights_path</span><span class="o">.</span><span class="n">encode</span><span class="p">(</span><span class="s2">"utf-8"</span><span class="p">),</span> <span class="mi">0</span> |
| <span class="p">)</span> |
| <span class="n">dshape</span> <span class="o">=</span> <span class="p">(</span><span class="n">env</span><span class="o">.</span><span class="n">BATCH</span><span class="p">,</span> <span class="n">net</span><span class="o">.</span><span class="n">c</span><span class="p">,</span> <span class="n">net</span><span class="o">.</span><span class="n">h</span><span class="p">,</span> <span class="n">net</span><span class="o">.</span><span class="n">w</span><span class="p">)</span> |
| <span class="n">dtype</span> <span class="o">=</span> <span class="s2">"float32"</span> |
| |
| <span class="c1"># Measure build start time</span> |
| <span class="n">build_start</span> <span class="o">=</span> <span class="n">time</span><span class="o">.</span><span class="n">time</span><span class="p">()</span> |
| |
| <span class="c1"># Start front end compilation</span> |
| <span class="n">mod</span><span class="p">,</span> <span class="n">params</span> <span class="o">=</span> <a href="../../../../reference/api/python/relay/frontend.html#tvm.relay.frontend.from_darknet" title="View documentation for tvm.relay.frontend.from_darknet"><span class="n">relay</span><span class="o">.</span><span class="n">frontend</span><span class="o">.</span><span class="n">from_darknet</span></a><span class="p">(</span><span class="n">net</span><span class="p">,</span> <span class="n">dtype</span><span class="o">=</span><span class="n">dtype</span><span class="p">,</span> <span class="n">shape</span><span class="o">=</span><span class="n">dshape</span><span class="p">)</span> |
| |
| <span class="k">if</span> <span class="n">target</span><span class="o">.</span><span class="n">device_name</span> <span class="o">==</span> <span class="s2">"vta"</span><span class="p">:</span> |
| <span class="c1"># Perform quantization in Relay</span> |
| <span class="c1"># Note: We set opt_level to 3 in order to fold batch norm</span> |
| <span class="k">with</span> <a href="../../../../reference/api/python/ir.html#tvm.transform.PassContext" title="View documentation for tvm.transform.PassContext"><span class="n">tvm</span><span class="o">.</span><span class="n">transform</span><span class="o">.</span><span class="n">PassContext</span></a><span class="p">(</span><span class="n">opt_level</span><span class="o">=</span><span class="mi">3</span><span class="p">):</span> |
| <span class="k">with</span> <span class="n">relay</span><span class="o">.</span><span class="n">quantize</span><span class="o">.</span><span class="n">qconfig</span><span class="p">(</span> |
| <span class="n">global_scale</span><span class="o">=</span><span class="mf">23.0</span><span class="p">,</span> |
| <span class="n">skip_conv_layers</span><span class="o">=</span><span class="p">[</span><span class="mi">0</span><span class="p">],</span> |
| <span class="n">store_lowbit_output</span><span class="o">=</span><span class="kc">True</span><span class="p">,</span> |
| <span class="n">round_for_shift</span><span class="o">=</span><span class="kc">True</span><span class="p">,</span> |
| <span class="p">):</span> |
| <span class="n">mod</span> <span class="o">=</span> <span class="n">relay</span><span class="o">.</span><span class="n">quantize</span><span class="o">.</span><span class="n">quantize</span><span class="p">(</span><span class="n">mod</span><span class="p">,</span> <span class="n">params</span><span class="o">=</span><span class="n">params</span><span class="p">)</span> |
| <span class="c1"># Perform graph packing and constant folding for VTA target</span> |
| <span class="n">mod</span> <span class="o">=</span> <span class="n">graph_pack</span><span class="p">(</span> |
| <span class="n">mod</span><span class="p">[</span><span class="s2">"main"</span><span class="p">],</span> |
| <span class="n">env</span><span class="o">.</span><span class="n">BATCH</span><span class="p">,</span> |
| <span class="n">env</span><span class="o">.</span><span class="n">BLOCK_OUT</span><span class="p">,</span> |
| <span class="n">env</span><span class="o">.</span><span class="n">WGT_WIDTH</span><span class="p">,</span> |
| <span class="n">start_name</span><span class="o">=</span><span class="n">pack_dict</span><span class="p">[</span><span class="n">MODEL_NAME</span><span class="p">][</span><span class="mi">0</span><span class="p">],</span> |
| <span class="n">stop_name</span><span class="o">=</span><span class="n">pack_dict</span><span class="p">[</span><span class="n">MODEL_NAME</span><span class="p">][</span><span class="mi">1</span><span class="p">],</span> |
| <span class="n">start_name_idx</span><span class="o">=</span><span class="n">pack_dict</span><span class="p">[</span><span class="n">MODEL_NAME</span><span class="p">][</span><span class="mi">2</span><span class="p">],</span> |
| <span class="n">stop_name_idx</span><span class="o">=</span><span class="n">pack_dict</span><span class="p">[</span><span class="n">MODEL_NAME</span><span class="p">][</span><span class="mi">3</span><span class="p">],</span> |
| <span class="p">)</span> |
| <span class="k">else</span><span class="p">:</span> |
| <span class="n">mod</span> <span class="o">=</span> <span class="n">mod</span><span class="p">[</span><span class="s2">"main"</span><span class="p">]</span> |
| |
| <span class="c1"># Compile Relay program with AlterOpLayout disabled</span> |
| <span class="k">with</span> <span class="n">vta</span><span class="o">.</span><span class="n">build_config</span><span class="p">(</span><span class="n">disabled_pass</span><span class="o">=</span><span class="p">{</span><span class="s2">"AlterOpLayout"</span><span class="p">,</span> <span class="s2">"tir.CommonSubexprElimTIR"</span><span class="p">}):</span> |
| <span class="n">lib</span> <span class="o">=</span> <span class="n">relay</span><span class="o">.</span><span class="n">build</span><span class="p">(</span> |
| <span class="n">mod</span><span class="p">,</span> <span class="n">target</span><span class="o">=</span><a href="../../../../reference/api/python/target.html#tvm.target.Target" title="View documentation for tvm.target.Target"><span class="n">tvm</span><span class="o">.</span><span class="n">target</span><span class="o">.</span><span class="n">Target</span></a><span class="p">(</span><span class="n">target</span><span class="p">,</span> <span class="n">host</span><span class="o">=</span><span class="n">env</span><span class="o">.</span><span class="n">target_host</span><span class="p">),</span> <span class="n">params</span><span class="o">=</span><span class="n">params</span> |
| <span class="p">)</span> |
| |
| <span class="c1"># Measure Relay build time</span> |
| <span class="n">build_time</span> <span class="o">=</span> <span class="n">time</span><span class="o">.</span><span class="n">time</span><span class="p">()</span> <span class="o">-</span> <span class="n">build_start</span> |
| <span class="nb">print</span><span class="p">(</span><span class="n">MODEL_NAME</span> <span class="o">+</span> <span class="s2">" inference graph built in </span><span class="si">{0:.2f}</span><span class="s2">s!"</span><span class="o">.</span><span class="n">format</span><span class="p">(</span><span class="n">build_time</span><span class="p">))</span> |
| |
| <span class="c1"># Send the inference library over to the remote RPC server</span> |
| <span class="n">temp</span> <span class="o">=</span> <a href="../../../../reference/api/python/contrib.html#tvm.contrib.utils.tempdir" title="View documentation for tvm.contrib.utils.tempdir"><span class="n">utils</span><span class="o">.</span><span class="n">tempdir</span></a><span class="p">()</span> |
| <span class="n">lib</span><span class="o">.</span><span class="n">export_library</span><span class="p">(</span><span class="n">temp</span><span class="o">.</span><span class="n">relpath</span><span class="p">(</span><span class="s2">"graphlib.tar"</span><span class="p">))</span> |
| <span class="n">remote</span><span class="o">.</span><span class="n">upload</span><span class="p">(</span><span class="n">temp</span><span class="o">.</span><span class="n">relpath</span><span class="p">(</span><span class="s2">"graphlib.tar"</span><span class="p">))</span> |
| <span class="n">lib</span> <span class="o">=</span> <span class="n">remote</span><span class="o">.</span><span class="n">load_module</span><span class="p">(</span><span class="s2">"graphlib.tar"</span><span class="p">)</span> |
| |
| <span class="c1"># Graph executor</span> |
| <span class="n">m</span> <span class="o">=</span> <a href="../../../../reference/api/python/graph_executor.html#tvm.contrib.graph_executor.GraphModule" title="View documentation for tvm.contrib.graph_executor.GraphModule"><span class="n">graph_executor</span><span class="o">.</span><span class="n">GraphModule</span></a><span class="p">(</span><span class="n">lib</span><span class="p">[</span><span class="s2">"default"</span><span class="p">](</span><span class="n">ctx</span><span class="p">))</span> |
| </pre></div> |
| </div> |
| <p class="sphx-glr-script-out">Out:</p> |
| <div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>/workspace/python/tvm/relay/build_module.py:439: DeprecationWarning: Please use input parameter mod (tvm.IRModule) instead of deprecated parameter mod (tvm.relay.function.Function) |
| DeprecationWarning, |
| yolov3-tiny inference graph built in 14.72s! |
| </pre></div> |
| </div> |
| </div> |
| <div class="section" id="perform-image-detection-inference"> |
| <h2>Perform image detection inference.<a class="headerlink" href="#perform-image-detection-inference" title="Permalink to this headline">¶</a></h2> |
| <p>We run detect on an downloaded image |
| Download test image</p> |
| <div class="highlight-default notranslate"><div class="highlight"><pre><span></span><span class="p">[</span><span class="n">neth</span><span class="p">,</span> <span class="n">netw</span><span class="p">]</span> <span class="o">=</span> <span class="n">dshape</span><span class="p">[</span><span class="mi">2</span><span class="p">:]</span> |
| <span class="n">test_image</span> <span class="o">=</span> <span class="s2">"person.jpg"</span> |
| <span class="n">img_url</span> <span class="o">=</span> <span class="n">REPO_URL</span> <span class="o">+</span> <span class="s2">"data/"</span> <span class="o">+</span> <span class="n">test_image</span> <span class="o">+</span> <span class="s2">"?raw=true"</span> |
| <span class="n">img_path</span> <span class="o">=</span> <span class="n">download_testdata</span><span class="p">(</span><span class="n">img_url</span><span class="p">,</span> <span class="n">test_image</span><span class="p">,</span> <span class="s2">"data"</span><span class="p">)</span> |
| <span class="n">data</span> <span class="o">=</span> <span class="n">darknet</span><span class="o">.</span><span class="n">load_image</span><span class="p">(</span><span class="n">img_path</span><span class="p">,</span> <span class="n">neth</span><span class="p">,</span> <span class="n">netw</span><span class="p">)</span><span class="o">.</span><span class="n">transpose</span><span class="p">(</span><span class="mi">1</span><span class="p">,</span> <span class="mi">2</span><span class="p">,</span> <span class="mi">0</span><span class="p">)</span> |
| |
| <span class="c1"># Prepare test image for inference</span> |
| <a href="https://matplotlib.org/stable/api/_as_gen/matplotlib.pyplot.imshow.html#matplotlib.pyplot.imshow" title="View documentation for matplotlib.pyplot.imshow"><span class="n">plt</span><span class="o">.</span><span class="n">imshow</span></a><span class="p">(</span><span class="n">data</span><span class="p">)</span> |
| <a href="https://matplotlib.org/stable/api/_as_gen/matplotlib.pyplot.show.html#matplotlib.pyplot.show" title="View documentation for matplotlib.pyplot.show"><span class="n">plt</span><span class="o">.</span><span class="n">show</span></a><span class="p">()</span> |
| <span class="n">data</span> <span class="o">=</span> <span class="n">data</span><span class="o">.</span><span class="n">transpose</span><span class="p">((</span><span class="mi">2</span><span class="p">,</span> <span class="mi">0</span><span class="p">,</span> <span class="mi">1</span><span class="p">))</span> |
| <span class="n">data</span> <span class="o">=</span> <span class="n">data</span><span class="p">[</span><a href="https://numpy.org/doc/stable/reference/constants.html#numpy.newaxis" title="View documentation for numpy.newaxis"><span class="n">np</span><span class="o">.</span><span class="n">newaxis</span></a><span class="p">,</span> <span class="p">:]</span> |
| <span class="n">data</span> <span class="o">=</span> <a href="https://numpy.org/doc/stable/reference/generated/numpy.repeat.html#numpy.repeat" title="View documentation for numpy.repeat"><span class="n">np</span><span class="o">.</span><span class="n">repeat</span></a><span class="p">(</span><span class="n">data</span><span class="p">,</span> <span class="n">env</span><span class="o">.</span><span class="n">BATCH</span><span class="p">,</span> <span class="n">axis</span><span class="o">=</span><span class="mi">0</span><span class="p">)</span> |
| |
| <span class="c1"># Set the network parameters and inputs</span> |
| <span class="n">m</span><span class="o">.</span><span class="n">set_input</span><span class="p">(</span><span class="s2">"data"</span><span class="p">,</span> <span class="n">data</span><span class="p">)</span> |
| |
| <span class="c1"># Perform inference and gather execution statistics</span> |
| <span class="c1"># More on: :py:method:`tvm.runtime.Module.time_evaluator`</span> |
| <span class="n">num</span> <span class="o">=</span> <span class="mi">4</span> <span class="c1"># number of times we run module for a single measurement</span> |
| <span class="n">rep</span> <span class="o">=</span> <span class="mi">3</span> <span class="c1"># number of measurements (we derive std dev from this)</span> |
| <span class="n">timer</span> <span class="o">=</span> <span class="n">m</span><span class="o">.</span><span class="n">module</span><span class="o">.</span><span class="n">time_evaluator</span><span class="p">(</span><span class="s2">"run"</span><span class="p">,</span> <span class="n">ctx</span><span class="p">,</span> <span class="n">number</span><span class="o">=</span><span class="n">num</span><span class="p">,</span> <span class="n">repeat</span><span class="o">=</span><span class="n">rep</span><span class="p">)</span> |
| |
| <span class="k">if</span> <span class="n">env</span><span class="o">.</span><span class="n">TARGET</span> <span class="ow">in</span> <span class="p">[</span><span class="s2">"sim"</span><span class="p">,</span> <span class="s2">"tsim"</span><span class="p">]:</span> |
| <span class="n">simulator</span><span class="o">.</span><span class="n">clear_stats</span><span class="p">()</span> |
| <span class="n">timer</span><span class="p">()</span> |
| <span class="n">sim_stats</span> <span class="o">=</span> <span class="n">simulator</span><span class="o">.</span><span class="n">stats</span><span class="p">()</span> |
| <span class="nb">print</span><span class="p">(</span><span class="s2">"</span><span class="se">\n</span><span class="s2">Execution statistics:"</span><span class="p">)</span> |
| <span class="k">for</span> <span class="n">k</span><span class="p">,</span> <span class="n">v</span> <span class="ow">in</span> <span class="n">sim_stats</span><span class="o">.</span><span class="n">items</span><span class="p">():</span> |
| <span class="c1"># Since we execute the workload many times, we need to normalize stats</span> |
| <span class="c1"># Note that there is always one warm up run</span> |
| <span class="c1"># Therefore we divide the overall stats by (num * rep + 1)</span> |
| <span class="nb">print</span><span class="p">(</span><span class="s2">"</span><span class="se">\t</span><span class="si">{:<16}</span><span class="s2">: </span><span class="si">{:>16}</span><span class="s2">"</span><span class="o">.</span><span class="n">format</span><span class="p">(</span><span class="n">k</span><span class="p">,</span> <span class="n">v</span> <span class="o">//</span> <span class="p">(</span><span class="n">num</span> <span class="o">*</span> <span class="n">rep</span> <span class="o">+</span> <span class="mi">1</span><span class="p">)))</span> |
| <span class="k">else</span><span class="p">:</span> |
| <span class="n">tcost</span> <span class="o">=</span> <span class="n">timer</span><span class="p">()</span> |
| <span class="n">std</span> <span class="o">=</span> <a href="https://numpy.org/doc/stable/reference/generated/numpy.std.html#numpy.std" title="View documentation for numpy.std"><span class="n">np</span><span class="o">.</span><span class="n">std</span></a><span class="p">(</span><span class="n">tcost</span><span class="o">.</span><span class="n">results</span><span class="p">)</span> <span class="o">*</span> <span class="mi">1000</span> |
| <span class="n">mean</span> <span class="o">=</span> <span class="n">tcost</span><span class="o">.</span><span class="n">mean</span> <span class="o">*</span> <span class="mi">1000</span> |
| <span class="nb">print</span><span class="p">(</span><span class="s2">"</span><span class="se">\n</span><span class="s2">Performed inference in </span><span class="si">%.2f</span><span class="s2">ms (std = </span><span class="si">%.2f</span><span class="s2">) for </span><span class="si">%d</span><span class="s2"> samples"</span> <span class="o">%</span> <span class="p">(</span><span class="n">mean</span><span class="p">,</span> <span class="n">std</span><span class="p">,</span> <span class="n">env</span><span class="o">.</span><span class="n">BATCH</span><span class="p">))</span> |
| <span class="nb">print</span><span class="p">(</span><span class="s2">"Average per sample inference time: </span><span class="si">%.2f</span><span class="s2">ms"</span> <span class="o">%</span> <span class="p">(</span><span class="n">mean</span> <span class="o">/</span> <span class="n">env</span><span class="o">.</span><span class="n">BATCH</span><span class="p">))</span> |
| |
| <span class="c1"># Get detection results from out</span> |
| <span class="n">thresh</span> <span class="o">=</span> <span class="mf">0.5</span> |
| <span class="n">nms_thresh</span> <span class="o">=</span> <span class="mf">0.45</span> |
| <span class="n">tvm_out</span> <span class="o">=</span> <span class="p">[]</span> |
| <span class="k">for</span> <span class="n">i</span> <span class="ow">in</span> <span class="nb">range</span><span class="p">(</span><span class="mi">2</span><span class="p">):</span> |
| <span class="n">layer_out</span> <span class="o">=</span> <span class="p">{}</span> |
| <span class="n">layer_out</span><span class="p">[</span><span class="s2">"type"</span><span class="p">]</span> <span class="o">=</span> <span class="s2">"Yolo"</span> |
| <span class="c1"># Get the yolo layer attributes (n, out_c, out_h, out_w, classes, total)</span> |
| <span class="n">layer_attr</span> <span class="o">=</span> <span class="n">m</span><span class="o">.</span><span class="n">get_output</span><span class="p">(</span><span class="n">i</span> <span class="o">*</span> <span class="mi">4</span> <span class="o">+</span> <span class="mi">3</span><span class="p">)</span><span class="o">.</span><span class="n">numpy</span><span class="p">()</span> |
| <span class="n">layer_out</span><span class="p">[</span><span class="s2">"biases"</span><span class="p">]</span> <span class="o">=</span> <span class="n">m</span><span class="o">.</span><span class="n">get_output</span><span class="p">(</span><span class="n">i</span> <span class="o">*</span> <span class="mi">4</span> <span class="o">+</span> <span class="mi">2</span><span class="p">)</span><span class="o">.</span><span class="n">numpy</span><span class="p">()</span> |
| <span class="n">layer_out</span><span class="p">[</span><span class="s2">"mask"</span><span class="p">]</span> <span class="o">=</span> <span class="n">m</span><span class="o">.</span><span class="n">get_output</span><span class="p">(</span><span class="n">i</span> <span class="o">*</span> <span class="mi">4</span> <span class="o">+</span> <span class="mi">1</span><span class="p">)</span><span class="o">.</span><span class="n">numpy</span><span class="p">()</span> |
| <span class="n">out_shape</span> <span class="o">=</span> <span class="p">(</span><span class="n">layer_attr</span><span class="p">[</span><span class="mi">0</span><span class="p">],</span> <span class="n">layer_attr</span><span class="p">[</span><span class="mi">1</span><span class="p">]</span> <span class="o">//</span> <span class="n">layer_attr</span><span class="p">[</span><span class="mi">0</span><span class="p">],</span> <span class="n">layer_attr</span><span class="p">[</span><span class="mi">2</span><span class="p">],</span> <span class="n">layer_attr</span><span class="p">[</span><span class="mi">3</span><span class="p">])</span> |
| <span class="n">layer_out</span><span class="p">[</span><span class="s2">"output"</span><span class="p">]</span> <span class="o">=</span> <span class="n">m</span><span class="o">.</span><span class="n">get_output</span><span class="p">(</span><span class="n">i</span> <span class="o">*</span> <span class="mi">4</span><span class="p">)</span><span class="o">.</span><span class="n">numpy</span><span class="p">()</span><span class="o">.</span><span class="n">reshape</span><span class="p">(</span><span class="n">out_shape</span><span class="p">)</span> |
| <span class="n">layer_out</span><span class="p">[</span><span class="s2">"classes"</span><span class="p">]</span> <span class="o">=</span> <span class="n">layer_attr</span><span class="p">[</span><span class="mi">4</span><span class="p">]</span> |
| <span class="n">tvm_out</span><span class="o">.</span><span class="n">append</span><span class="p">(</span><span class="n">layer_out</span><span class="p">)</span> |
| <span class="n">thresh</span> <span class="o">=</span> <span class="mf">0.560</span> |
| |
| <span class="c1"># Show detection results</span> |
| <span class="n">img</span> <span class="o">=</span> <span class="n">darknet</span><span class="o">.</span><span class="n">load_image_color</span><span class="p">(</span><span class="n">img_path</span><span class="p">)</span> |
| <span class="n">_</span><span class="p">,</span> <span class="n">im_h</span><span class="p">,</span> <span class="n">im_w</span> <span class="o">=</span> <span class="n">img</span><span class="o">.</span><span class="n">shape</span> |
| <span class="n">dets</span> <span class="o">=</span> <span class="n">yolo_detection</span><span class="o">.</span><span class="n">fill_network_boxes</span><span class="p">((</span><span class="n">netw</span><span class="p">,</span> <span class="n">neth</span><span class="p">),</span> <span class="p">(</span><span class="n">im_w</span><span class="p">,</span> <span class="n">im_h</span><span class="p">),</span> <span class="n">thresh</span><span class="p">,</span> <span class="mi">1</span><span class="p">,</span> <span class="n">tvm_out</span><span class="p">)</span> |
| <span class="n">last_layer</span> <span class="o">=</span> <span class="n">net</span><span class="o">.</span><span class="n">layers</span><span class="p">[</span><span class="n">net</span><span class="o">.</span><span class="n">n</span> <span class="o">-</span> <span class="mi">1</span><span class="p">]</span> |
| <span class="n">yolo_detection</span><span class="o">.</span><span class="n">do_nms_sort</span><span class="p">(</span><span class="n">dets</span><span class="p">,</span> <span class="n">last_layer</span><span class="o">.</span><span class="n">classes</span><span class="p">,</span> <span class="n">nms_thresh</span><span class="p">)</span> |
| <span class="n">yolo_detection</span><span class="o">.</span><span class="n">draw_detections</span><span class="p">(</span><span class="n">font_path</span><span class="p">,</span> <span class="n">img</span><span class="p">,</span> <span class="n">dets</span><span class="p">,</span> <span class="n">thresh</span><span class="p">,</span> <span class="n">names</span><span class="p">,</span> <span class="n">last_layer</span><span class="o">.</span><span class="n">classes</span><span class="p">)</span> |
| <a href="https://matplotlib.org/stable/api/_as_gen/matplotlib.pyplot.imshow.html#matplotlib.pyplot.imshow" title="View documentation for matplotlib.pyplot.imshow"><span class="n">plt</span><span class="o">.</span><span class="n">imshow</span></a><span class="p">(</span><span class="n">img</span><span class="o">.</span><span class="n">transpose</span><span class="p">(</span><span class="mi">1</span><span class="p">,</span> <span class="mi">2</span><span class="p">,</span> <span class="mi">0</span><span class="p">))</span> |
| <a href="https://matplotlib.org/stable/api/_as_gen/matplotlib.pyplot.show.html#matplotlib.pyplot.show" title="View documentation for matplotlib.pyplot.show"><span class="n">plt</span><span class="o">.</span><span class="n">show</span></a><span class="p">()</span> |
| </pre></div> |
| </div> |
| <img alt="../../../../_images/sphx_glr_deploy_detection_001.png" class="sphx-glr-single-img" src="../../../../_images/sphx_glr_deploy_detection_001.png" /> |
| <p class="sphx-glr-script-out">Out:</p> |
| <div class="sphx-glr-script-out highlight-none notranslate"><div class="highlight"><pre><span></span>Execution statistics: |
| inp_load_nbytes : 25462784 |
| wgt_load_nbytes : 17558016 |
| acc_load_nbytes : 96128 |
| uop_load_nbytes : 5024 |
| out_store_nbytes: 3396224 |
| gemm_counter : 10578048 |
| alu_counter : 849056 |
| </pre></div> |
| </div> |
| <div class="sphx-glr-footer class sphx-glr-footer-example docutils container" id="sphx-glr-download-topic-vta-tutorials-frontend-deploy-detection-py"> |
| <div class="sphx-glr-download docutils container"> |
| <p><a class="reference download internal" download="" href="../../../../_downloads/65b9451c8de050d7cd9da2fe5a49acc6/deploy_detection.py"><code class="xref download docutils literal notranslate"><span class="pre">Download</span> <span class="pre">Python</span> <span class="pre">source</span> <span class="pre">code:</span> <span class="pre">deploy_detection.py</span></code></a></p> |
| </div> |
| <div class="sphx-glr-download docutils container"> |
| <p><a class="reference download internal" download="" href="../../../../_downloads/66e1a42229aae7ed49ac268f520e6727/deploy_detection.ipynb"><code class="xref download docutils literal notranslate"><span class="pre">Download</span> <span class="pre">Jupyter</span> <span class="pre">notebook:</span> <span class="pre">deploy_detection.ipynb</span></code></a></p> |
| </div> |
| </div> |
| <p class="sphx-glr-signature"><a class="reference external" href="https://sphinx-gallery.github.io">Gallery generated by Sphinx-Gallery</a></p> |
| </div> |
| </div> |
| |
| |
| </div> |
| |
| </div> |
| |
| |
| <footer> |
| |
| <div class="rst-footer-buttons" role="navigation" aria-label="footer navigation"> |
| |
| <a href="../optimize/convolution_opt.html" class="btn btn-neutral float-right" title="2D Convolution Optimization" accesskey="n" rel="next">Next <span class="fa fa-arrow-circle-right"></span></a> |
| |
| |
| <a href="deploy_classification.html" class="btn btn-neutral float-left" title="Deploy Pretrained Vision Model from MxNet on VTA" accesskey="p" rel="prev"><span class="fa fa-arrow-circle-left"></span> Previous</a> |
| |
| </div> |
| |
| <div id="button" class="backtop"><img src="../../../../_static//img/right.svg" alt="backtop"/> </div> |
| <section class="footerSec"> |
| <div class="footerHeader"> |
| <ul class="d-flex align-md-items-center justify-content-between flex-column flex-md-row"> |
| <li class="copywrite d-flex align-items-center"> |
| <h5 id="copy-right-info">© 2020 Apache Software Foundation | All right reserved</h5> |
| </li> |
| </ul> |
| |
| </div> |
| |
| <ul> |
| <li class="footernote">Copyright © 2020 The Apache Software Foundation. Apache TVM, Apache, the Apache feather, and the Apache TVM project logo are either trademarks or registered trademarks of the Apache Software Foundation.</li> |
| </ul> |
| |
| </section> |
| </footer> |
| </div> |
| </div> |
| |
| </section> |
| |
| </div> |
| |
| |
| <script src="https://cdnjs.cloudflare.com/ajax/libs/popper.js/1.12.9/umd/popper.min.js" integrity="sha384-ApNbgh9B+Y1QKtv3Rn7W3mgPxhU9K/ScQsAP7hUibX39j7fakFPskvXusvfa0b4Q" crossorigin="anonymous"></script> |
| <script src="https://maxcdn.bootstrapcdn.com/bootstrap/4.0.0/js/bootstrap.min.js" integrity="sha384-JZR6Spejh4U02d8jOt6vLEHfe/JQGiRRSQQxSfFWpi1MquVdAyjUar5+76PVCmYl" crossorigin="anonymous"></script> |
| |
| </body> |
| <script type="text/javascript"> |
| jQuery(function () { |
| SphinxRtdTheme.Navigation.enable(true); |
| }); |
| </script> |
| |
| |
| |
| |
| <!-- Theme Analytics --> |
| <script> |
| (function(i,s,o,g,r,a,m){i['GoogleAnalyticsObject']=r;i[r]=i[r]||function(){ |
| (i[r].q=i[r].q||[]).push(arguments)},i[r].l=1*new Date();a=s.createElement(o), |
| m=s.getElementsByTagName(o)[0];a.async=1;a.src=g;m.parentNode.insertBefore(a,m) |
| })(window,document,'script','https://www.google-analytics.com/analytics.js','ga'); |
| |
| ga('create', 'UA-75982049-2', 'auto'); |
| ga('send', 'pageview'); |
| </script> |
| |
| |
| |
| |
| </body> |
| </html> |
