341 lines
9.7 KiB
HTML
341 lines
9.7 KiB
HTML
|
|
|||
|
|
|
|||
|
|
<!DOCTYPE html>
|
|||
|
|
<!--[if IE 8]><html class="no-js lt-ie9" lang="en" > <![endif]-->
|
|||
|
|
<!--[if gt IE 8]><!--> <html class="no-js" lang="en" > <!--<![endif]-->
|
|||
|
|
<head>
|
|||
|
|
<meta charset="utf-8">
|
|||
|
|
|
|||
|
|
<meta name="viewport" content="width=device-width, initial-scale=1.0">
|
|||
|
|
|
|||
|
|
<title>Benchmarking 3D — Point Cloud Library 1.12.0 documentation</title>
|
|||
|
|
|
|||
|
|
|
|||
|
|
|
|||
|
|
|
|||
|
|
|
|||
|
|
|
|||
|
|
|
|||
|
|
|
|||
|
|
<script type="text/javascript" src="_static/js/modernizr.min.js"></script>
|
|||
|
|
|
|||
|
|
|
|||
|
|
<script type="text/javascript" id="documentation_options" data-url_root="./" src="_static/documentation_options.js"></script>
|
|||
|
|
<script type="text/javascript" src="_static/jquery.js"></script>
|
|||
|
|
<script type="text/javascript" src="_static/underscore.js"></script>
|
|||
|
|
<script type="text/javascript" src="_static/doctools.js"></script>
|
|||
|
|
<script type="text/javascript" src="_static/language_data.js"></script>
|
|||
|
|
|
|||
|
|
<script type="text/javascript" src="_static/js/theme.js"></script>
|
|||
|
|
|
|||
|
|
|
|||
|
|
|
|||
|
|
|
|||
|
|
<link rel="stylesheet" href="_static/css/theme.css" type="text/css" />
|
|||
|
|
<link rel="stylesheet" href="_static/pygments.css" type="text/css" />
|
|||
|
|
<link rel="index" title="Index" href="genindex.html" />
|
|||
|
|
<link rel="search" title="Search" href="search.html" />
|
|||
|
|
</head>
|
|||
|
|
|
|||
|
|
<body class="wy-body-for-nav">
|
|||
|
|
|
|||
|
|
|
|||
|
|
<div class="wy-grid-for-nav">
|
|||
|
|
|
|||
|
|
<nav data-toggle="wy-nav-shift" class="wy-nav-side">
|
|||
|
|
<div class="wy-side-scroll">
|
|||
|
|
<div class="wy-side-nav-search" >
|
|||
|
|
|
|||
|
|
|
|||
|
|
|
|||
|
|
<a href="index.html" class="icon icon-home"> Point Cloud Library
|
|||
|
|
|
|||
|
|
|
|||
|
|
|
|||
|
|
</a>
|
|||
|
|
|
|||
|
|
|
|||
|
|
|
|||
|
|
|
|||
|
|
<div class="version">
|
|||
|
|
1.12.0
|
|||
|
|
</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">
|
|||
|
|
|
|||
|
|
|
|||
|
|
|
|||
|
|
|
|||
|
|
|
|||
|
|
|
|||
|
|
<!-- Local TOC -->
|
|||
|
|
<div class="local-toc"><ul>
|
|||
|
|
<li><a class="reference internal" href="#">Benchmarking 3D</a></li>
|
|||
|
|
<li><a class="reference internal" href="#benchmarking-object-recognition">Benchmarking Object Recognition</a><ul>
|
|||
|
|
<li><a class="reference internal" href="#training">1. Training</a></li>
|
|||
|
|
<li><a class="reference internal" href="#keypoints">2. Keypoints</a></li>
|
|||
|
|
<li><a class="reference internal" href="#descriptors">3. Descriptors</a></li>
|
|||
|
|
<li><a class="reference internal" href="#classification">4. Classification</a></li>
|
|||
|
|
<li><a class="reference internal" href="#evaluation">5. Evaluation</a><ul>
|
|||
|
|
<li><a class="reference internal" href="#metrics">5.1 Metrics</a><ul>
|
|||
|
|
<li><a class="reference internal" href="#object-recognition-api">Object Recognition API</a></li>
|
|||
|
|
</ul>
|
|||
|
|
</li>
|
|||
|
|
</ul>
|
|||
|
|
</li>
|
|||
|
|
<li><a class="reference internal" href="#id1">1. Training</a></li>
|
|||
|
|
<li><a class="reference internal" href="#id2">2. Keypoints</a></li>
|
|||
|
|
<li><a class="reference internal" href="#id3">3. Descriptors</a></li>
|
|||
|
|
<li><a class="reference internal" href="#id4">4. Classification</a></li>
|
|||
|
|
<li><a class="reference internal" href="#id5">5. Evaluation</a></li>
|
|||
|
|
</ul>
|
|||
|
|
</li>
|
|||
|
|
</ul>
|
|||
|
|
</div>
|
|||
|
|
|
|||
|
|
|
|||
|
|
</div>
|
|||
|
|
</div>
|
|||
|
|
</nav>
|
|||
|
|
|
|||
|
|
<section data-toggle="wy-nav-shift" class="wy-nav-content-wrap">
|
|||
|
|
|
|||
|
|
|
|||
|
|
<nav class="wy-nav-top" aria-label="top navigation">
|
|||
|
|
|
|||
|
|
<i data-toggle="wy-nav-top" class="fa fa-bars"></i>
|
|||
|
|
<a href="index.html">Point Cloud Library</a>
|
|||
|
|
|
|||
|
|
</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> »</li>
|
|||
|
|
|
|||
|
|
<li>Benchmarking 3D</li>
|
|||
|
|
|
|||
|
|
|
|||
|
|
<li class="wy-breadcrumbs-aside">
|
|||
|
|
|
|||
|
|
|
|||
|
|
|
|||
|
|
</li>
|
|||
|
|
|
|||
|
|
</ul>
|
|||
|
|
|
|||
|
|
|
|||
|
|
<hr/>
|
|||
|
|
</div>
|
|||
|
|
<div role="main" class="document" itemscope="itemscope" itemtype="http://schema.org/Article">
|
|||
|
|
<div itemprop="articleBody">
|
|||
|
|
|
|||
|
|
<div class="section" id="benchmarking-3d">
|
|||
|
|
<span id="benchmarking"></span><h1>Benchmarking 3D</h1>
|
|||
|
|
<p>This document introduces benchmarking concepts for 3D algorithms. By
|
|||
|
|
<em>benchmarking</em> here we refer to the possibility of testing different
|
|||
|
|
computational pipelines in an <strong>easy manner</strong>. The goal is to test their
|
|||
|
|
reproductibility with respect to a particular problem of general interest.</p>
|
|||
|
|
</div>
|
|||
|
|
<div class="section" id="benchmarking-object-recognition">
|
|||
|
|
<h1>Benchmarking Object Recognition</h1>
|
|||
|
|
<p>For the general problem of Object Recognition (identification, categorization,
|
|||
|
|
detection, etc – all fall in the same category here), we identify the
|
|||
|
|
following steps:</p>
|
|||
|
|
<div class="section" id="training">
|
|||
|
|
<h2>1. Training</h2>
|
|||
|
|
<p>Users should be able to acquire training data from different inputs, including
|
|||
|
|
but not limited to:</p>
|
|||
|
|
<blockquote>
|
|||
|
|
<div><ul class="simple">
|
|||
|
|
<li>full triangle meshes (CAD models);</li>
|
|||
|
|
<li>360-degree full point cloud models;</li>
|
|||
|
|
<li>partial point cloud views:<ul>
|
|||
|
|
<li>in clutter;</li>
|
|||
|
|
<li>cleanly segmented.</li>
|
|||
|
|
</ul>
|
|||
|
|
</li>
|
|||
|
|
</ul>
|
|||
|
|
</div></blockquote>
|
|||
|
|
</div>
|
|||
|
|
<div class="section" id="keypoints">
|
|||
|
|
<h2>2. Keypoints</h2>
|
|||
|
|
<p>Computing higher level representation from the object’s appearance (texture + depth) should be done:</p>
|
|||
|
|
<blockquote>
|
|||
|
|
<div><ul class="simple">
|
|||
|
|
<li><strong>densely</strong> - at every point/vertex in the input data;</li>
|
|||
|
|
<li>at certain <strong>interest points</strong> (i.e., keypoints).</li>
|
|||
|
|
</ul>
|
|||
|
|
</div></blockquote>
|
|||
|
|
<p>The detected keypoint might also contain some meta-information required by some descriptors, like scale or orientation.</p>
|
|||
|
|
</div>
|
|||
|
|
<div class="section" id="descriptors">
|
|||
|
|
<h2>3. Descriptors</h2>
|
|||
|
|
<p>A higher level representation as mentioned before will be herein represented by a <strong>feature descriptor</strong>. Feature descriptors can be:</p>
|
|||
|
|
<blockquote>
|
|||
|
|
<div><ul class="simple">
|
|||
|
|
<li>2D (two-dimensional) – here we refer to those descriptors estimated solely from RGB texture data;</li>
|
|||
|
|
<li>3D (three-dimensional) – here we refer to those descriptors estimated solely from XYZ/depth data;</li>
|
|||
|
|
<li>a combination of the above.</li>
|
|||
|
|
</ul>
|
|||
|
|
</div></blockquote>
|
|||
|
|
<p>In addition, feature descriptors can be:</p>
|
|||
|
|
<blockquote>
|
|||
|
|
<div><ul class="simple">
|
|||
|
|
<li><strong>local</strong> - estimated only at a set of discrete keypoints, using the information from neighboring pixels/points;</li>
|
|||
|
|
<li><strong>global</strong>, or meta-local - estimated on entire objects or the entire input dataset.</li>
|
|||
|
|
</ul>
|
|||
|
|
</div></blockquote>
|
|||
|
|
</div>
|
|||
|
|
<div class="section" id="classification">
|
|||
|
|
<h2>4. Classification</h2>
|
|||
|
|
<p>The distribution of features should be classifiable into distinct, separable
|
|||
|
|
classes. For local features, we identify two sets of techniques:</p>
|
|||
|
|
<blockquote>
|
|||
|
|
<div><ul class="simple">
|
|||
|
|
<li><strong>bag of words</strong>;</li>
|
|||
|
|
<li><strong>voting</strong>;</li>
|
|||
|
|
<li><strong>supervised voting</strong> (regression from the description to the relative 3D location, e.g. Hough forest).</li>
|
|||
|
|
</ul>
|
|||
|
|
</div></blockquote>
|
|||
|
|
<p>For global features, any general purpose classification technique should work (e.g., SVMs, nearest neighbors, etc).</p>
|
|||
|
|
<p>In addition to classification, a substep of it could be considered
|
|||
|
|
<strong>Registration</strong>. Here we refine the classification results using iterative
|
|||
|
|
closest point techniques for example.</p>
|
|||
|
|
</div>
|
|||
|
|
<div class="section" id="evaluation">
|
|||
|
|
<h2>5. Evaluation</h2>
|
|||
|
|
<p>This pipeline should be able to evaluate the algorithm’s performance at
|
|||
|
|
different tasks. Here are some requested tasks to support:</p>
|
|||
|
|
<blockquote>
|
|||
|
|
<div><ul class="simple">
|
|||
|
|
<li>object id and pose</li>
|
|||
|
|
<li>object id and segmentation</li>
|
|||
|
|
<li>object id and bounding box</li>
|
|||
|
|
<li>category and segmentation</li>
|
|||
|
|
<li>category and bounding box</li>
|
|||
|
|
</ul>
|
|||
|
|
</div></blockquote>
|
|||
|
|
<div class="section" id="metrics">
|
|||
|
|
<h3>5.1 Metrics</h3>
|
|||
|
|
<p>This pipeline should provide different metrics, since algorithms excel in
|
|||
|
|
different areas. Here are some requested metrics:</p>
|
|||
|
|
<blockquote>
|
|||
|
|
<div><ul class="simple">
|
|||
|
|
<li>precision-recall</li>
|
|||
|
|
<li>time</li>
|
|||
|
|
<li>average rank of correct id</li>
|
|||
|
|
<li>area under curve of cumulative histogram of rank of correct id</li>
|
|||
|
|
</ul>
|
|||
|
|
</div></blockquote>
|
|||
|
|
<div class="section" id="object-recognition-api">
|
|||
|
|
<h4>Object Recognition API</h4>
|
|||
|
|
<p>Here we describe a proposed set of classes that could be easily extended and
|
|||
|
|
used for the purpose of benchmarking object recognition tasks.</p>
|
|||
|
|
</div>
|
|||
|
|
</div>
|
|||
|
|
</div>
|
|||
|
|
<div class="section" id="id1">
|
|||
|
|
<h2>1. Training</h2>
|
|||
|
|
</div>
|
|||
|
|
<div class="section" id="id2">
|
|||
|
|
<h2>2. Keypoints</h2>
|
|||
|
|
</div>
|
|||
|
|
<div class="section" id="id3">
|
|||
|
|
<h2>3. Descriptors</h2>
|
|||
|
|
</div>
|
|||
|
|
<div class="section" id="id4">
|
|||
|
|
<h2>4. Classification</h2>
|
|||
|
|
</div>
|
|||
|
|
<div class="section" id="id5">
|
|||
|
|
<h2>5. Evaluation</h2>
|
|||
|
|
<p>The evaluation output needs to be one of the following:</p>
|
|||
|
|
<blockquote>
|
|||
|
|
<div><ul class="simple">
|
|||
|
|
<li>object id</li>
|
|||
|
|
<li>object pose</li>
|
|||
|
|
<li>object category</li>
|
|||
|
|
<li>object bounding box</li>
|
|||
|
|
<li>object mask</li>
|
|||
|
|
</ul>
|
|||
|
|
</div></blockquote>
|
|||
|
|
</div>
|
|||
|
|
</div>
|
|||
|
|
|
|||
|
|
|
|||
|
|
</div>
|
|||
|
|
|
|||
|
|
</div>
|
|||
|
|
<footer>
|
|||
|
|
|
|||
|
|
|
|||
|
|
<hr/>
|
|||
|
|
|
|||
|
|
<div role="contentinfo">
|
|||
|
|
<p>
|
|||
|
|
© Copyright
|
|||
|
|
|
|||
|
|
</p>
|
|||
|
|
</div>
|
|||
|
|
Built with <a href="http://sphinx-doc.org/">Sphinx</a> using a <a href="https://github.com/rtfd/sphinx_rtd_theme">theme</a> provided by <a href="https://readthedocs.org">Read the Docs</a>.
|
|||
|
|
|
|||
|
|
</footer>
|
|||
|
|
|
|||
|
|
</div>
|
|||
|
|
</div>
|
|||
|
|
|
|||
|
|
</section>
|
|||
|
|
|
|||
|
|
</div>
|
|||
|
|
|
|||
|
|
|
|||
|
|
|
|||
|
|
<script type="text/javascript">
|
|||
|
|
jQuery(function () {
|
|||
|
|
SphinxRtdTheme.Navigation.enable(true);
|
|||
|
|
});
|
|||
|
|
</script>
|
|||
|
|
|
|||
|
|
|
|||
|
|
|
|||
|
|
|
|||
|
|
|
|||
|
|
|
|||
|
|
</body>
|
|||
|
|
</html>
|