Google Git
Sign in
apache / mxnet-site / 79788b4935d532765935d81e0b76a5d2258b7907 / . / api / python / ndarray / contrib.html
blob: b52efac0c0ebfe929b36857023a466a803e30c60 [file] [log] [blame]
<!DOCTYPE html>
<html lang="en">
<head>
<meta charset="utf-8"/>
<meta content="IE=edge" http-equiv="X-UA-Compatible"/>
<meta content="width=device-width, initial-scale=1" name="viewport"/>
<title>Contrib NDArray API — mxnet documentation</title>
<link crossorigin="anonymous" href="https://maxcdn.bootstrapcdn.com/bootstrap/3.3.6/css/bootstrap.min.css" integrity="sha384-1q8mTJOASx8j1Au+a5WDVnPi2lkFfwwEAa8hDDdjZlpLegxhjVME1fgjWPGmkzs7" rel="stylesheet"/>
<link href="https://maxcdn.bootstrapcdn.com/font-awesome/4.5.0/css/font-awesome.min.css" rel="stylesheet"/>
<link href="../../../_static/basic.css" rel="stylesheet" type="text/css">
<link href="../../../_static/pygments.css" rel="stylesheet" type="text/css">
<link href="../../../_static/mxnet.css" rel="stylesheet" type="text/css"/>
<script type="text/javascript">
var DOCUMENTATION_OPTIONS = {
URL_ROOT: '../../../',
VERSION: '',
COLLAPSE_INDEX: false,
FILE_SUFFIX: '.html',
HAS_SOURCE: true,
SOURCELINK_SUFFIX: ''
};
</script>
<script src="https://code.jquery.com/jquery-1.11.1.min.js" type="text/javascript"></script>
<script src="../../../_static/underscore.js" type="text/javascript"></script>
<script src="../../../_static/searchtools_custom.js" type="text/javascript"></script>
<script src="../../../_static/doctools.js" type="text/javascript"></script>
<script src="../../../_static/selectlang.js" type="text/javascript"></script>
<script src="https://cdnjs.cloudflare.com/ajax/libs/mathjax/2.7.1/MathJax.js?config=TeX-AMS-MML_HTMLorMML" type="text/javascript"></script>
<script type="text/javascript"> jQuery(function() { Search.loadIndex("/searchindex.js"); Search.init();}); </script>
<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-96378503-1', 'auto');
ga('send', 'pageview');
</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="https://cdn.mathjax.org/mathjax/latest/MathJax.js?config=TeX-AMS-MML_HTMLorMML"></script> -->
<!-- -->
<link href="../index.html" rel="up" title="MXNet - Python API">
<link href="../symbol/symbol.html" rel="next" title="Symbol API"/>
<link href="sparse.html" rel="prev" title="Sparse NDArray API"/>
<link href="https://raw.githubusercontent.com/dmlc/web-data/master/mxnet/image/mxnet-icon.png" rel="icon" type="image/png"/>
</link></link></link></head>
<body background="https://raw.githubusercontent.com/dmlc/web-data/master/mxnet/image/mxnet-background-compressed.jpeg" role="document">
<div class="content-block"><div class="navbar navbar-fixed-top">
<div class="container" id="navContainer">
<div class="innder" id="header-inner">
<h1 id="logo-wrap">
<a href="../../../" id="logo"><img src="https://raw.githubusercontent.com/dmlc/web-data/master/mxnet/image/mxnet_logo.png"/></a>
</h1>
<nav class="nav-bar" id="main-nav">
<a class="main-nav-link" href="../../../install/index.html">Install</a>
<span id="dropdown-menu-position-anchor">
<a aria-expanded="true" aria-haspopup="true" class="main-nav-link dropdown-toggle" data-toggle="dropdown" href="#" role="button">Gluon <span class="caret"></span></a>
<ul class="dropdown-menu navbar-menu" id="package-dropdown-menu">
<li><a class="main-nav-link" href="../../../gluon/index.html">About</a></li>
<li><a class="main-nav-link" href="http://gluon.mxnet.io">Tutorials</a></li>
</ul>
</span>
<span id="dropdown-menu-position-anchor">
<a aria-expanded="true" aria-haspopup="true" class="main-nav-link dropdown-toggle" data-toggle="dropdown" href="#" role="button">API <span class="caret"></span></a>
<ul class="dropdown-menu navbar-menu" id="package-dropdown-menu">
<li><a class="main-nav-link" href="../../../api/python/index.html">Python</a></li>
<li><a class="main-nav-link" href="../../../api/scala/index.html">Scala</a></li>
<li><a class="main-nav-link" href="../../../api/r/index.html">R</a></li>
<li><a class="main-nav-link" href="../../../api/julia/index.html">Julia</a></li>
<li><a class="main-nav-link" href="../../../api/c++/index.html">C++</a></li>
<li><a class="main-nav-link" href="../../../api/perl/index.html">Perl</a></li>
</ul>
</span>
<span id="dropdown-menu-position-anchor-docs">
<a aria-expanded="true" aria-haspopup="true" class="main-nav-link dropdown-toggle" data-toggle="dropdown" href="#" role="button">Docs <span class="caret"></span></a>
<ul class="dropdown-menu navbar-menu" id="package-dropdown-menu-docs">
<li><a class="main-nav-link" href="../../../tutorials/index.html">Tutorials</a>
<li><a class="main-nav-link" href="../../../faq/index.html">FAQ</a></li>
<li><a class="main-nav-link" href="../../../architecture/index.html">Architecture</a></li>
<li><a class="main-nav-link" href="https://github.com/apache/incubator-mxnet/tree/1.1.0/example">Examples</a></li>
<li><a class="main-nav-link" href="../../../api/python/gluon/model_zoo.html">Gluon Model Zoo</a></li>
</li></ul>
</span>
<a class="main-nav-link" href="https://github.com/dmlc/mxnet">Github</a>
<span id="dropdown-menu-position-anchor-community">
<a aria-expanded="true" aria-haspopup="true" class="main-nav-link dropdown-toggle" data-toggle="dropdown" href="#" role="button">Community <span class="caret"></span></a>
<ul class="dropdown-menu navbar-menu" id="package-dropdown-menu-community">
<li><a class="main-nav-link" href="../../../community/index.html">Community</a></li>
<li><a class="main-nav-link" href="../../../community/contribute.html">Contribute</a></li>
<li><a class="main-nav-link" href="../../../community/powered_by.html">Powered By</a></li>
<li><a class="main-nav-link" href="http://discuss.mxnet.io">Discuss</a></li>
</ul>
</span>
<span id="dropdown-menu-position-anchor-version" style="position: relative"><a href="#" class="main-nav-link dropdown-toggle" data-toggle="dropdown" role="button" aria-haspopup="true" aria-expanded="true">Versions(1.1.0)<span class="caret"></span></a><ul id="package-dropdown-menu" class="dropdown-menu"><li><a class="main-nav-link" href=https://mxnet.incubator.apache.org/>1.1.0</a></li><li><a class="main-nav-link" href=https://mxnet.incubator.apache.org/versions/1.0.0/index.html>1.0.0</a></li><li><a class="main-nav-link" href=https://mxnet.incubator.apache.org/versions/0.12.1/index.html>0.12.1</a></li><li><a class="main-nav-link" href=https://mxnet.incubator.apache.org/versions/0.12.0/index.html>0.12.0</a></li><li><a class="main-nav-link" href=https://mxnet.incubator.apache.org/versions/0.11.0/index.html>0.11.0</a></li><li><a class="main-nav-link" href=https://mxnet.incubator.apache.org/versions/master/index.html>master</a></li></ul></span></nav>
<script> function getRootPath(){ return "../../../" } </script>
<div class="burgerIcon dropdown">
<a class="dropdown-toggle" data-toggle="dropdown" href="#" role="button">☰</a>
<ul class="dropdown-menu" id="burgerMenu">
<li><a href="../../../install/index.html">Install</a></li>
<li><a class="main-nav-link" href="../../../tutorials/index.html">Tutorials</a></li>
<li class="dropdown-submenu">
<a href="#" tabindex="-1">Community</a>
<ul class="dropdown-menu">
<li><a href="../../../community/index.html" tabindex="-1">Community</a></li>
<li><a href="../../../community/contribute.html" tabindex="-1">Contribute</a></li>
<li><a href="../../../community/powered_by.html" tabindex="-1">Powered By</a></li>
</ul>
</li>
<li class="dropdown-submenu">
<a href="#" tabindex="-1">API</a>
<ul class="dropdown-menu">
<li><a href="../../../api/python/index.html" tabindex="-1">Python</a>
</li>
<li><a href="../../../api/scala/index.html" tabindex="-1">Scala</a>
</li>
<li><a href="../../../api/r/index.html" tabindex="-1">R</a>
</li>
<li><a href="../../../api/julia/index.html" tabindex="-1">Julia</a>
</li>
<li><a href="../../../api/c++/index.html" tabindex="-1">C++</a>
</li>
<li><a href="../../../api/perl/index.html" tabindex="-1">Perl</a>
</li>
</ul>
</li>
<li class="dropdown-submenu">
<a href="#" tabindex="-1">Docs</a>
<ul class="dropdown-menu">
<li><a href="../../../tutorials/index.html" tabindex="-1">Tutorials</a></li>
<li><a href="../../../faq/index.html" tabindex="-1">FAQ</a></li>
<li><a href="../../../architecture/index.html" tabindex="-1">Architecture</a></li>
<li><a href="https://github.com/apache/incubator-mxnet/tree/1.1.0/example" tabindex="-1">Examples</a></li>
<li><a href="../../../api/python/gluon/model_zoo.html" tabindex="-1">Gluon Model Zoo</a></li>
</ul>
</li>
<li><a href="../../../architecture/index.html">Architecture</a></li>
<li><a class="main-nav-link" href="https://github.com/dmlc/mxnet">Github</a></li>
<li id="dropdown-menu-position-anchor-version-mobile" class="dropdown-submenu" style="position: relative"><a href="#" tabindex="-1">Versions(1.1.0)</a><ul class="dropdown-menu"><li><a tabindex="-1" href=https://mxnet.incubator.apache.org/>1.1.0</a></li><li><a tabindex="-1" href=https://mxnet.incubator.apache.org/versions/1.0.0/index.html>1.0.0</a></li><li><a tabindex="-1" href=https://mxnet.incubator.apache.org/versions/0.12.1/index.html>0.12.1</a></li><li><a tabindex="-1" href=https://mxnet.incubator.apache.org/versions/0.12.0/index.html>0.12.0</a></li><li><a tabindex="-1" href=https://mxnet.incubator.apache.org/versions/0.11.0/index.html>0.11.0</a></li><li><a tabindex="-1" href=https://mxnet.incubator.apache.org/versions/master/index.html>master</a></li></ul></li></ul>
</div>
<div class="plusIcon dropdown">
<a class="dropdown-toggle" data-toggle="dropdown" href="#" role="button"><span aria-hidden="true" class="glyphicon glyphicon-plus"></span></a>
<ul class="dropdown-menu dropdown-menu-right" id="plusMenu"></ul>
</div>
<div id="search-input-wrap">
<form action="../../../search.html" autocomplete="off" class="" method="get" role="search">
<div class="form-group inner-addon left-addon">
<i class="glyphicon glyphicon-search"></i>
<input class="form-control" name="q" placeholder="Search" type="text"/>
</div>
<input name="check_keywords" type="hidden" value="yes">
<input name="area" type="hidden" value="default"/>
</input></form>
<div id="search-preview"></div>
</div>
<div id="searchIcon">
<span aria-hidden="true" class="glyphicon glyphicon-search"></span>
</div>
<!-- <div id="lang-select-wrap"> -->
<!-- <label id="lang-select-label"> -->
<!-- <\!-- <i class="fa fa-globe"></i> -\-> -->
<!-- <span></span> -->
<!-- </label> -->
<!-- <select id="lang-select"> -->
<!-- <option value="en">Eng</option> -->
<!-- <option value="zh">中文</option> -->
<!-- </select> -->
<!-- </div> -->
<!-- <a id="mobile-nav-toggle">
<span class="mobile-nav-toggle-bar"></span>
<span class="mobile-nav-toggle-bar"></span>
<span class="mobile-nav-toggle-bar"></span>
</a> -->
</div>
</div>
</div>
<script type="text/javascript">
$('body').css('background', 'white');
</script>
<div class="container">
<div class="row">
<div aria-label="main navigation" class="sphinxsidebar leftsidebar" role="navigation">
<div class="sphinxsidebarwrapper">
<ul class="current">
<li class="toctree-l1 current"><a class="reference internal" href="../index.html">Python Documents</a><ul class="current">
<li class="toctree-l2 current"><a class="reference internal" href="../index.html#ndarray-api">NDArray API</a><ul class="current">
<li class="toctree-l3"><a class="reference internal" href="ndarray.html">NDArray API</a></li>
<li class="toctree-l3"><a class="reference internal" href="random.html">Random Distribution Generator NDArray API</a></li>
<li class="toctree-l3"><a class="reference internal" href="linalg.html">Linear Algebra NDArray API</a></li>
<li class="toctree-l3"><a class="reference internal" href="sparse.html">Sparse NDArray API</a></li>
<li class="toctree-l3 current"><a class="current reference internal" href="">Contrib NDArray API</a><ul>
<li class="toctree-l4"><a class="reference internal" href="#overview">Overview</a></li>
<li class="toctree-l4"><a class="reference internal" href="#contrib">Contrib</a></li>
<li class="toctree-l4"><a class="reference internal" href="#api-reference">API Reference</a></li>
</ul>
</li>
</ul>
</li>
<li class="toctree-l2"><a class="reference internal" href="../index.html#symbol-api">Symbol API</a></li>
<li class="toctree-l2"><a class="reference internal" href="../index.html#module-api">Module API</a></li>
<li class="toctree-l2"><a class="reference internal" href="../index.html#autograd-api">Autograd API</a></li>
<li class="toctree-l2"><a class="reference internal" href="../index.html#gluon-api">Gluon API</a></li>
<li class="toctree-l2"><a class="reference internal" href="../index.html#kvstore-api">KVStore API</a></li>
<li class="toctree-l2"><a class="reference internal" href="../index.html#io-api">IO API</a></li>
<li class="toctree-l2"><a class="reference internal" href="../index.html#image-api">Image API</a></li>
<li class="toctree-l2"><a class="reference internal" href="../index.html#optimization-api">Optimization API</a></li>
<li class="toctree-l2"><a class="reference internal" href="../index.html#callback-api">Callback API</a></li>
<li class="toctree-l2"><a class="reference internal" href="../index.html#metric-api">Metric API</a></li>
<li class="toctree-l2"><a class="reference internal" href="../index.html#run-time-compilation-api">Run-Time Compilation API</a></li>
<li class="toctree-l2"><a class="reference internal" href="../index.html#contrib-package">Contrib Package</a></li>
</ul>
</li>
<li class="toctree-l1"><a class="reference internal" href="../../r/index.html">R Documents</a></li>
<li class="toctree-l1"><a class="reference internal" href="../../julia/index.html">Julia Documents</a></li>
<li class="toctree-l1"><a class="reference internal" href="../../c++/index.html">C++ Documents</a></li>
<li class="toctree-l1"><a class="reference internal" href="../../scala/index.html">Scala Documents</a></li>
<li class="toctree-l1"><a class="reference internal" href="../../perl/index.html">Perl Documents</a></li>
<li class="toctree-l1"><a class="reference internal" href="../../../faq/index.html">HowTo Documents</a></li>
<li class="toctree-l1"><a class="reference internal" href="../../../architecture/index.html">System Documents</a></li>
<li class="toctree-l1"><a class="reference internal" href="../../../tutorials/index.html">Tutorials</a></li>
<li class="toctree-l1"><a class="reference internal" href="../../../community/index.html">Community</a></li>
</ul>
</div>
</div>
<div class="content">
<div class="page-tracker"></div>
<div class="section" id="contrib-ndarray-api">
<span id="contrib-ndarray-api"></span><h1>Contrib NDArray API<a class="headerlink" href="#contrib-ndarray-api" title="Permalink to this headline"></a></h1>
<div class="section" id="overview">
<span id="overview"></span><h2>Overview<a class="headerlink" href="#overview" title="Permalink to this headline"></a></h2>
<p>This document lists the contrib routines of the <em>n</em>-dimensional array package:</p>
<table border="1" class="longtable docutils">
<colgroup>
<col width="10%"/>
<col width="90%"/>
</colgroup>
<tbody valign="top">
<tr class="row-odd"><td><a class="reference internal" href="#module-mxnet.ndarray.contrib" title="mxnet.ndarray.contrib"><code class="xref py py-obj docutils literal"><span class="pre">mxnet.ndarray.contrib</span></code></a></td>
<td>Contrib NDArray API of MXNet.</td>
</tr>
</tbody>
</table>
<p>The <code class="docutils literal"><span class="pre">Contrib</span> <span class="pre">NDArray</span></code> API, defined in the <code class="docutils literal"><span class="pre">ndarray.contrib</span></code> package, provides
many useful experimental APIs for new features.
This is a place for the community to try out the new features,
so that feature contributors can receive feedback.</p>
<div class="admonition warning">
<p class="first admonition-title">Warning</p>
<p class="last">This package contains experimental APIs and may change in the near future.</p>
</div>
<p>In the rest of this document, we list routines provided by the <code class="docutils literal"><span class="pre">ndarray.contrib</span></code> package.</p>
</div>
<div class="section" id="contrib">
<span id="contrib"></span><h2>Contrib<a class="headerlink" href="#contrib" title="Permalink to this headline"></a></h2>
<table border="1" class="longtable docutils">
<colgroup>
<col width="10%"/>
<col width="90%"/>
</colgroup>
<tbody valign="top">
<tr class="row-odd"><td><a class="reference internal" href="#mxnet.ndarray.contrib.CTCLoss" title="mxnet.ndarray.contrib.CTCLoss"><code class="xref py py-obj docutils literal"><span class="pre">CTCLoss</span></code></a></td>
<td>Connectionist Temporal Classification Loss.</td>
</tr>
<tr class="row-even"><td><a class="reference internal" href="#mxnet.ndarray.contrib.DeformableConvolution" title="mxnet.ndarray.contrib.DeformableConvolution"><code class="xref py py-obj docutils literal"><span class="pre">DeformableConvolution</span></code></a></td>
<td>Compute 2-D deformable convolution on 4-D input.</td>
</tr>
<tr class="row-odd"><td><a class="reference internal" href="#mxnet.ndarray.contrib.DeformablePSROIPooling" title="mxnet.ndarray.contrib.DeformablePSROIPooling"><code class="xref py py-obj docutils literal"><span class="pre">DeformablePSROIPooling</span></code></a></td>
<td>Performs deformable position-sensitive region-of-interest pooling on inputs.</td>
</tr>
<tr class="row-even"><td><a class="reference internal" href="#mxnet.ndarray.contrib.MultiBoxDetection" title="mxnet.ndarray.contrib.MultiBoxDetection"><code class="xref py py-obj docutils literal"><span class="pre">MultiBoxDetection</span></code></a></td>
<td>Convert multibox detection predictions.</td>
</tr>
<tr class="row-odd"><td><a class="reference internal" href="#mxnet.ndarray.contrib.MultiBoxPrior" title="mxnet.ndarray.contrib.MultiBoxPrior"><code class="xref py py-obj docutils literal"><span class="pre">MultiBoxPrior</span></code></a></td>
<td>Generate prior(anchor) boxes from data, sizes and ratios.</td>
</tr>
<tr class="row-even"><td><a class="reference internal" href="#mxnet.ndarray.contrib.MultiBoxTarget" title="mxnet.ndarray.contrib.MultiBoxTarget"><code class="xref py py-obj docutils literal"><span class="pre">MultiBoxTarget</span></code></a></td>
<td>Compute Multibox training targets</td>
</tr>
<tr class="row-odd"><td><a class="reference internal" href="#mxnet.ndarray.contrib.MultiProposal" title="mxnet.ndarray.contrib.MultiProposal"><code class="xref py py-obj docutils literal"><span class="pre">MultiProposal</span></code></a></td>
<td>Generate region proposals via RPN</td>
</tr>
<tr class="row-even"><td><a class="reference internal" href="#mxnet.ndarray.contrib.PSROIPooling" title="mxnet.ndarray.contrib.PSROIPooling"><code class="xref py py-obj docutils literal"><span class="pre">PSROIPooling</span></code></a></td>
<td>Performs region-of-interest pooling on inputs.</td>
</tr>
<tr class="row-odd"><td><a class="reference internal" href="#mxnet.ndarray.contrib.Proposal" title="mxnet.ndarray.contrib.Proposal"><code class="xref py py-obj docutils literal"><span class="pre">Proposal</span></code></a></td>
<td>Generate region proposals via RPN</td>
</tr>
<tr class="row-even"><td><a class="reference internal" href="#mxnet.ndarray.contrib.count_sketch" title="mxnet.ndarray.contrib.count_sketch"><code class="xref py py-obj docutils literal"><span class="pre">count_sketch</span></code></a></td>
<td>Apply CountSketch to input: map a d-dimension data to k-dimension data”</td>
</tr>
<tr class="row-odd"><td><a class="reference internal" href="#mxnet.ndarray.contrib.ctc_loss" title="mxnet.ndarray.contrib.ctc_loss"><code class="xref py py-obj docutils literal"><span class="pre">ctc_loss</span></code></a></td>
<td>Connectionist Temporal Classification Loss.</td>
</tr>
<tr class="row-even"><td><a class="reference internal" href="#mxnet.ndarray.contrib.dequantize" title="mxnet.ndarray.contrib.dequantize"><code class="xref py py-obj docutils literal"><span class="pre">dequantize</span></code></a></td>
<td>Dequantize the input tensor into a float tensor.</td>
</tr>
<tr class="row-odd"><td><a class="reference internal" href="#mxnet.ndarray.contrib.fft" title="mxnet.ndarray.contrib.fft"><code class="xref py py-obj docutils literal"><span class="pre">fft</span></code></a></td>
<td>Apply 1D FFT to input”</td>
</tr>
<tr class="row-even"><td><a class="reference internal" href="#mxnet.ndarray.contrib.ifft" title="mxnet.ndarray.contrib.ifft"><code class="xref py py-obj docutils literal"><span class="pre">ifft</span></code></a></td>
<td>Apply 1D ifft to input”</td>
</tr>
<tr class="row-odd"><td><a class="reference internal" href="#mxnet.ndarray.contrib.quantize" title="mxnet.ndarray.contrib.quantize"><code class="xref py py-obj docutils literal"><span class="pre">quantize</span></code></a></td>
<td>Quantize a input tensor from float to <cite>out_type</cite>, with user-specified <cite>min_range</cite> and <cite>max_range</cite>.</td>
</tr>
</tbody>
</table>
</div>
<div class="section" id="api-reference">
<span id="api-reference"></span><h2>API Reference<a class="headerlink" href="#api-reference" title="Permalink to this headline"></a></h2>
<script src="../../../_static/js/auto_module_index.js" type="text/javascript"></script><span class="target" id="module-mxnet.ndarray.contrib"></span><p>Contrib NDArray API of MXNet.</p>
<dl class="function">
<dt id="mxnet.ndarray.contrib.CTCLoss">
<code class="descclassname">mxnet.ndarray.contrib.</code><code class="descname">CTCLoss</code><span class="sig-paren">(</span><em>data=None</em>, <em>label=None</em>, <em>data_lengths=None</em>, <em>label_lengths=None</em>, <em>use_data_lengths=_Null</em>, <em>use_label_lengths=_Null</em>, <em>blank_label=_Null</em>, <em>out=None</em>, <em>name=None</em>, <em>**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.ndarray.contrib.CTCLoss" title="Permalink to this definition"></a></dt>
<dd><p>Connectionist Temporal Classification Loss.</p>
<p>The shapes of the inputs and outputs:</p>
<ul class="simple">
<li><strong>data</strong>: <cite>(sequence_length, batch_size, alphabet_size)</cite></li>
<li><strong>label</strong>: <cite>(batch_size, label_sequence_length)</cite></li>
<li><strong>out</strong>: <cite>(batch_size)</cite></li>
</ul>
<p>The <cite>data</cite> tensor consists of sequences of activation vectors (without applying softmax),
with i-th channel in the last dimension corresponding to i-th label
for i between 0 and alphabet_size-1 (i.e always 0-indexed).
Alphabet size should include one additional value reserved for blank label.
When <cite>blank_label</cite> is <code class="docutils literal"><span class="pre">"first"</span></code>, the <code class="docutils literal"><span class="pre">0</span></code>-th channel is be reserved for
activation of blank label, or otherwise if it is “last”, <code class="docutils literal"><span class="pre">(alphabet_size-1)</span></code>-th channel should be
reserved for blank label.</p>
<p><code class="docutils literal"><span class="pre">label</span></code> is an index matrix of integers. When <cite>blank_label</cite> is <code class="docutils literal"><span class="pre">"first"</span></code>,
the value 0 is then reserved for blank label, and should not be passed in this matrix. Otherwise,
when <cite>blank_label</cite> is <code class="docutils literal"><span class="pre">"last"</span></code>, the value <cite>(alphabet_size-1)</cite> is reserved for blank label.</p>
<p>If a sequence of labels is shorter than <em>label_sequence_length</em>, use the special
padding value at the end of the sequence to conform it to the correct
length. The padding value is <cite>0</cite> when <cite>blank_label</cite> is <code class="docutils literal"><span class="pre">"first"</span></code>, and <cite>-1</cite> otherwise.</p>
<p>For example, suppose the vocabulary is <cite>[a, b, c]</cite>, and in one batch we have three sequences
‘ba’, ‘cbb’, and ‘abac’. When <cite>blank_label</cite> is <code class="docutils literal"><span class="pre">"first"</span></code>, we can index the labels as
<cite>{‘a’: 1, ‘b’: 2, ‘c’: 3}</cite>, and we reserve the 0-th channel for blank label in data tensor.
The resulting <cite>label</cite> tensor should be padded to be:</p>
<div class="highlight-python"><div class="highlight"><pre><span></span><span class="p">[[</span><span class="mi">2</span><span class="p">,</span> <span class="mi">1</span><span class="p">,</span> <span class="mi">0</span><span class="p">,</span> <span class="mi">0</span><span class="p">],</span> <span class="p">[</span><span class="mi">3</span><span class="p">,</span> <span class="mi">2</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="p">[</span><span class="mi">1</span><span class="p">,</span> <span class="mi">2</span><span class="p">,</span> <span class="mi">1</span><span class="p">,</span> <span class="mi">3</span><span class="p">]]</span>
</pre></div>
</div>
<p>When <cite>blank_label</cite> is <code class="docutils literal"><span class="pre">"last"</span></code>, we can index the labels as
<cite>{‘a’: 0, ‘b’: 1, ‘c’: 2}</cite>, and we reserve the channel index 3 for blank label in data tensor.
The resulting <cite>label</cite> tensor should be padded to be:</p>
<div class="highlight-python"><div class="highlight"><pre><span></span><span class="p">[[</span><span class="mi">1</span><span class="p">,</span> <span class="mi">0</span><span class="p">,</span> <span class="o">-</span><span class="mi">1</span><span class="p">,</span> <span class="o">-</span><span class="mi">1</span><span class="p">],</span> <span class="p">[</span><span class="mi">2</span><span class="p">,</span> <span class="mi">1</span><span class="p">,</span> <span class="mi">1</span><span class="p">,</span> <span class="o">-</span><span class="mi">1</span><span class="p">],</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="mi">0</span><span class="p">,</span> <span class="mi">2</span><span class="p">]]</span>
</pre></div>
</div>
<p><code class="docutils literal"><span class="pre">out</span></code> is a list of CTC loss values, one per example in the batch.</p>
<p>See <em>Connectionist Temporal Classification: Labelling Unsegmented
Sequence Data with Recurrent Neural Networks</em>, A. Graves <em>et al</em>. for more
information on the definition and the algorithm.</p>
<p>Defined in src/operator/contrib/ctc_loss.cc:L115</p>
<table class="docutils field-list" frame="void" rules="none">
<col class="field-name"/>
<col class="field-body"/>
<tbody valign="top">
<tr class="field-odd field"><th class="field-name">Parameters:</th><td class="field-body"><ul class="first simple">
<li><strong>data</strong> (<a class="reference internal" href="ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – Input data to the ctc_loss op.</li>
<li><strong>label</strong> (<a class="reference internal" href="ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – Ground-truth labels for the loss.</li>
<li><strong>data_lengths</strong> (<a class="reference internal" href="ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – Lengths of data for each of the samples. Only required when use_data_lengths is true.</li>
<li><strong>label_lengths</strong> (<a class="reference internal" href="ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – Lengths of labels for each of the samples. Only required when use_label_lengths is true.</li>
<li><strong>use_data_lengths</strong> (<em>boolean, optional, default=0</em>) – Whether the data lenghts are decided by <cite>data_lengths</cite>. If false, the lengths are equal to the max sequence length.</li>
<li><strong>use_label_lengths</strong> (<em>boolean, optional, default=0</em>) – Whether the label lenghts are decided by <cite>label_lengths</cite>, or derived from <cite>padding_mask</cite>. If false, the lengths are derived from the first occurrence of the value of <cite>padding_mask</cite>. The value of <cite>padding_mask</cite> is <code class="docutils literal"><span class="pre">0</span></code> when first CTC label is reserved for blank, and <code class="docutils literal"><span class="pre">-1</span></code> when last label is reserved for blank. See <cite>blank_label</cite>.</li>
<li><strong>blank_label</strong> (<em>{'first', 'last'},optional, default='first'</em>) – Set the label that is reserved for blank label.If “first”, 0-th label is reserved, and label values for tokens in the vocabulary are between <code class="docutils literal"><span class="pre">1</span></code> and <code class="docutils literal"><span class="pre">alphabet_size-1</span></code>, and the padding mask is <code class="docutils literal"><span class="pre">-1</span></code>. If “last”, last label value <code class="docutils literal"><span class="pre">alphabet_size-1</span></code> is reserved for blank label instead, and label values for tokens in the vocabulary are between <code class="docutils literal"><span class="pre">0</span></code> and <code class="docutils literal"><span class="pre">alphabet_size-2</span></code>, and the padding mask is <code class="docutils literal"><span class="pre">0</span></code>.</li>
<li><strong>out</strong> (<em>NDArray, optional</em>) – The output NDArray to hold the result.</li>
</ul>
</td>
</tr>
<tr class="field-even field"><th class="field-name">Returns:</th><td class="field-body"><p class="first"><strong>out</strong> –
The output of this function.</p>
</td>
</tr>
<tr class="field-odd field"><th class="field-name">Return type:</th><td class="field-body"><p class="first last">NDArray or list of NDArrays</p>
</td>
</tr>
</tbody>
</table>
</dd></dl>
<dl class="function">
<dt id="mxnet.ndarray.contrib.DeformableConvolution">
<code class="descclassname">mxnet.ndarray.contrib.</code><code class="descname">DeformableConvolution</code><span class="sig-paren">(</span><em>data=None</em>, <em>offset=None</em>, <em>weight=None</em>, <em>bias=None</em>, <em>kernel=_Null</em>, <em>stride=_Null</em>, <em>dilate=_Null</em>, <em>pad=_Null</em>, <em>num_filter=_Null</em>, <em>num_group=_Null</em>, <em>num_deformable_group=_Null</em>, <em>workspace=_Null</em>, <em>no_bias=_Null</em>, <em>layout=_Null</em>, <em>out=None</em>, <em>name=None</em>, <em>**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.ndarray.contrib.DeformableConvolution" title="Permalink to this definition"></a></dt>
<dd><p>Compute 2-D deformable convolution on 4-D input.</p>
<p>The deformable convolution operation is described in <a class="reference external" href="https://arxiv.org/abs/1703.06211">https://arxiv.org/abs/1703.06211</a></p>
<p>For 2-D deformable convolution, the shapes are</p>
<ul class="simple">
<li><strong>data</strong>: <em>(batch_size, channel, height, width)</em></li>
<li><strong>offset</strong>: <em>(batch_size, num_deformable_group * kernel[0] * kernel[1], height, width)</em></li>
<li><strong>weight</strong>: <em>(num_filter, channel, kernel[0], kernel[1])</em></li>
<li><strong>bias</strong>: <em>(num_filter,)</em></li>
<li><strong>out</strong>: <em>(batch_size, num_filter, out_height, out_width)</em>.</li>
</ul>
<p>Define:</p>
<div class="highlight-python"><div class="highlight"><pre><span></span><span class="n">f</span><span class="p">(</span><span class="n">x</span><span class="p">,</span><span class="n">k</span><span class="p">,</span><span class="n">p</span><span class="p">,</span><span class="n">s</span><span class="p">,</span><span class="n">d</span><span class="p">)</span> <span class="o">=</span> <span class="n">floor</span><span class="p">((</span><span class="n">x</span><span class="o">+</span><span class="mi">2</span><span class="o">*</span><span class="n">p</span><span class="o">-</span><span class="n">d</span><span class="o">*</span><span class="p">(</span><span class="n">k</span><span class="o">-</span><span class="mi">1</span><span class="p">)</span><span class="o">-</span><span class="mi">1</span><span class="p">)</span><span class="o">/</span><span class="n">s</span><span class="p">)</span><span class="o">+</span><span class="mi">1</span>
</pre></div>
</div>
<p>then we have:</p>
<div class="highlight-python"><div class="highlight"><pre><span></span><span class="n">out_height</span><span class="o">=</span><span class="n">f</span><span class="p">(</span><span class="n">height</span><span class="p">,</span> <span class="n">kernel</span><span class="p">[</span><span class="mi">0</span><span class="p">],</span> <span class="n">pad</span><span class="p">[</span><span class="mi">0</span><span class="p">],</span> <span class="n">stride</span><span class="p">[</span><span class="mi">0</span><span class="p">],</span> <span class="n">dilate</span><span class="p">[</span><span class="mi">0</span><span class="p">])</span>
<span class="n">out_width</span><span class="o">=</span><span class="n">f</span><span class="p">(</span><span class="n">width</span><span class="p">,</span> <span class="n">kernel</span><span class="p">[</span><span class="mi">1</span><span class="p">],</span> <span class="n">pad</span><span class="p">[</span><span class="mi">1</span><span class="p">],</span> <span class="n">stride</span><span class="p">[</span><span class="mi">1</span><span class="p">],</span> <span class="n">dilate</span><span class="p">[</span><span class="mi">1</span><span class="p">])</span>
</pre></div>
</div>
<p>If <code class="docutils literal"><span class="pre">no_bias</span></code> is set to be true, then the <code class="docutils literal"><span class="pre">bias</span></code> term is ignored.</p>
<p>The default data <code class="docutils literal"><span class="pre">layout</span></code> is <em>NCHW</em>, namely <em>(batch_size, channle, height,
width)</em>.</p>
<p>If <code class="docutils literal"><span class="pre">num_group</span></code> is larger than 1, denoted by <em>g</em>, then split the input <code class="docutils literal"><span class="pre">data</span></code>
evenly into <em>g</em> parts along the channel axis, and also evenly split <code class="docutils literal"><span class="pre">weight</span></code>
along the first dimension. Next compute the convolution on the <em>i</em>-th part of
the data with the <em>i</em>-th weight part. The output is obtained by concating all
the <em>g</em> results.</p>
<p>If <code class="docutils literal"><span class="pre">num_deformable_group</span></code> is larger than 1, denoted by <em>dg</em>, then split the
input <code class="docutils literal"><span class="pre">offset</span></code> evenly into <em>dg</em> parts along the channel axis, and also evenly
split <code class="docutils literal"><span class="pre">out</span></code> evenly into <em>dg</em> parts along the channel axis. Next compute the
deformable convolution, apply the <em>i</em>-th part of the offset part on the <em>i</em>-th
out.</p>
<p>Both <code class="docutils literal"><span class="pre">weight</span></code> and <code class="docutils literal"><span class="pre">bias</span></code> are learnable parameters.</p>
<p>Defined in src/operator/contrib/deformable_convolution.cc:L100</p>
<table class="docutils field-list" frame="void" rules="none">
<col class="field-name"/>
<col class="field-body"/>
<tbody valign="top">
<tr class="field-odd field"><th class="field-name">Parameters:</th><td class="field-body"><ul class="first simple">
<li><strong>data</strong> (<a class="reference internal" href="ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – Input data to the DeformableConvolutionOp.</li>
<li><strong>offset</strong> (<a class="reference internal" href="ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – Input offset to the DeformableConvolutionOp.</li>
<li><strong>weight</strong> (<a class="reference internal" href="ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – Weight matrix.</li>
<li><strong>bias</strong> (<a class="reference internal" href="ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – Bias parameter.</li>
<li><strong>kernel</strong> (<em>Shape(tuple), required</em>) – Convolution kernel size: (h, w) or (d, h, w)</li>
<li><strong>stride</strong> (<em>Shape(tuple), optional, default=[]</em>) – Convolution stride: (h, w) or (d, h, w). Defaults to 1 for each dimension.</li>
<li><strong>dilate</strong> (<em>Shape(tuple), optional, default=[]</em>) – Convolution dilate: (h, w) or (d, h, w). Defaults to 1 for each dimension.</li>
<li><strong>pad</strong> (<em>Shape(tuple), optional, default=[]</em>) – Zero pad for convolution: (h, w) or (d, h, w). Defaults to no padding.</li>
<li><strong>num_filter</strong> (<em>int (non-negative), required</em>) – Convolution filter(channel) number</li>
<li><strong>num_group</strong> (<em>int (non-negative), optional, default=1</em>) – Number of group partitions.</li>
<li><strong>num_deformable_group</strong> (<em>int (non-negative), optional, default=1</em>) – Number of deformable group partitions.</li>
<li><strong>workspace</strong> (<em>long (non-negative), optional, default=1024</em>) – Maximum temperal workspace allowed for convolution (MB).</li>
<li><strong>no_bias</strong> (<em>boolean, optional, default=0</em>) – Whether to disable bias parameter.</li>
<li><strong>layout</strong> (<em>{None, 'NCDHW', 'NCHW', 'NCW'},optional, default='None'</em>) – Set layout for input, output and weight. Empty for
default layout: NCW for 1d, NCHW for 2d and NCDHW for 3d.</li>
<li><strong>out</strong> (<em>NDArray, optional</em>) – The output NDArray to hold the result.</li>
</ul>
</td>
</tr>
<tr class="field-even field"><th class="field-name">Returns:</th><td class="field-body"><p class="first"><strong>out</strong> –
The output of this function.</p>
</td>
</tr>
<tr class="field-odd field"><th class="field-name">Return type:</th><td class="field-body"><p class="first last">NDArray or list of NDArrays</p>
</td>
</tr>
</tbody>
</table>
</dd></dl>
<dl class="function">
<dt id="mxnet.ndarray.contrib.DeformablePSROIPooling">
<code class="descclassname">mxnet.ndarray.contrib.</code><code class="descname">DeformablePSROIPooling</code><span class="sig-paren">(</span><em>data=None</em>, <em>rois=None</em>, <em>trans=None</em>, <em>spatial_scale=_Null</em>, <em>output_dim=_Null</em>, <em>group_size=_Null</em>, <em>pooled_size=_Null</em>, <em>part_size=_Null</em>, <em>sample_per_part=_Null</em>, <em>trans_std=_Null</em>, <em>no_trans=_Null</em>, <em>out=None</em>, <em>name=None</em>, <em>**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.ndarray.contrib.DeformablePSROIPooling" title="Permalink to this definition"></a></dt>
<dd><p>Performs deformable position-sensitive region-of-interest pooling on inputs.
The DeformablePSROIPooling operation is described in <a class="reference external" href="https://arxiv.org/abs/1703.06211">https://arxiv.org/abs/1703.06211</a> .batch_size will change to the number of region bounding boxes after DeformablePSROIPooling</p>
<table class="docutils field-list" frame="void" rules="none">
<col class="field-name"/>
<col class="field-body"/>
<tbody valign="top">
<tr class="field-odd field"><th class="field-name">Parameters:</th><td class="field-body"><ul class="first simple">
<li><strong>data</strong> (<a class="reference internal" href="../symbol/symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Input data to the pooling operator, a 4D Feature maps</li>
<li><strong>rois</strong> (<a class="reference internal" href="../symbol/symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Bounding box coordinates, a 2D array of [[batch_index, x1, y1, x2, y2]]. (x1, y1) and (x2, y2) are top left and down right corners of designated region of interest. batch_index indicates the index of corresponding image in the input data</li>
<li><strong>trans</strong> (<a class="reference internal" href="../symbol/symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – transition parameter</li>
<li><strong>spatial_scale</strong> (<em>float, required</em>) – Ratio of input feature map height (or w) to raw image height (or w). Equals the reciprocal of total stride in convolutional layers</li>
<li><strong>output_dim</strong> (<em>int, required</em>) – fix output dim</li>
<li><strong>group_size</strong> (<em>int, required</em>) – fix group size</li>
<li><strong>pooled_size</strong> (<em>int, required</em>) – fix pooled size</li>
<li><strong>part_size</strong> (<em>int, optional, default='0'</em>) – fix part size</li>
<li><strong>sample_per_part</strong> (<em>int, optional, default='1'</em>) – fix samples per part</li>
<li><strong>trans_std</strong> (<em>float, optional, default=0</em>) – fix transition std</li>
<li><strong>no_trans</strong> (<em>boolean, optional, default=0</em>) – Whether to disable trans parameter.</li>
<li><strong>out</strong> (<em>NDArray, optional</em>) – The output NDArray to hold the result.</li>
</ul>
</td>
</tr>
<tr class="field-even field"><th class="field-name">Returns:</th><td class="field-body"><p class="first"><strong>out</strong> –
The output of this function.</p>
</td>
</tr>
<tr class="field-odd field"><th class="field-name">Return type:</th><td class="field-body"><p class="first last">NDArray or list of NDArrays</p>
</td>
</tr>
</tbody>
</table>
</dd></dl>
<dl class="function">
<dt id="mxnet.ndarray.contrib.MultiBoxDetection">
<code class="descclassname">mxnet.ndarray.contrib.</code><code class="descname">MultiBoxDetection</code><span class="sig-paren">(</span><em>cls_prob=None</em>, <em>loc_pred=None</em>, <em>anchor=None</em>, <em>clip=_Null</em>, <em>threshold=_Null</em>, <em>background_id=_Null</em>, <em>nms_threshold=_Null</em>, <em>force_suppress=_Null</em>, <em>variances=_Null</em>, <em>nms_topk=_Null</em>, <em>out=None</em>, <em>name=None</em>, <em>**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.ndarray.contrib.MultiBoxDetection" title="Permalink to this definition"></a></dt>
<dd><p>Convert multibox detection predictions.</p>
<table class="docutils field-list" frame="void" rules="none">
<col class="field-name"/>
<col class="field-body"/>
<tbody valign="top">
<tr class="field-odd field"><th class="field-name">Parameters:</th><td class="field-body"><ul class="first simple">
<li><strong>cls_prob</strong> (<a class="reference internal" href="ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – Class probabilities.</li>
<li><strong>loc_pred</strong> (<a class="reference internal" href="ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – Location regression predictions.</li>
<li><strong>anchor</strong> (<a class="reference internal" href="ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – Multibox prior anchor boxes</li>
<li><strong>clip</strong> (<em>boolean, optional, default=1</em>) – Clip out-of-boundary boxes.</li>
<li><strong>threshold</strong> (<em>float, optional, default=0.01</em>) – Threshold to be a positive prediction.</li>
<li><strong>background_id</strong> (<em>int, optional, default='0'</em>) – Background id.</li>
<li><strong>nms_threshold</strong> (<em>float, optional, default=0.5</em>) – Non-maximum suppression threshold.</li>
<li><strong>force_suppress</strong> (<em>boolean, optional, default=0</em>) – Suppress all detections regardless of class_id.</li>
<li><strong>variances</strong> (<em>tuple of <float>, optional, default=[0.1,0.1,0.2,0.2]</em>) – Variances to be decoded from box regression output.</li>
<li><strong>nms_topk</strong> (<em>int, optional, default='-1'</em>) – Keep maximum top k detections before nms, -1 for no limit.</li>
<li><strong>out</strong> (<em>NDArray, optional</em>) – The output NDArray to hold the result.</li>
</ul>
</td>
</tr>
<tr class="field-even field"><th class="field-name">Returns:</th><td class="field-body"><p class="first"><strong>out</strong> –
The output of this function.</p>
</td>
</tr>
<tr class="field-odd field"><th class="field-name">Return type:</th><td class="field-body"><p class="first last">NDArray or list of NDArrays</p>
</td>
</tr>
</tbody>
</table>
</dd></dl>
<dl class="function">
<dt id="mxnet.ndarray.contrib.MultiBoxPrior">
<code class="descclassname">mxnet.ndarray.contrib.</code><code class="descname">MultiBoxPrior</code><span class="sig-paren">(</span><em>data=None</em>, <em>sizes=_Null</em>, <em>ratios=_Null</em>, <em>clip=_Null</em>, <em>steps=_Null</em>, <em>offsets=_Null</em>, <em>out=None</em>, <em>name=None</em>, <em>**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.ndarray.contrib.MultiBoxPrior" title="Permalink to this definition"></a></dt>
<dd><p>Generate prior(anchor) boxes from data, sizes and ratios.</p>
<table class="docutils field-list" frame="void" rules="none">
<col class="field-name"/>
<col class="field-body"/>
<tbody valign="top">
<tr class="field-odd field"><th class="field-name">Parameters:</th><td class="field-body"><ul class="first simple">
<li><strong>data</strong> (<a class="reference internal" href="ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – Input data.</li>
<li><strong>sizes</strong> (<em>tuple of <float>, optional, default=[1]</em>) – List of sizes of generated MultiBoxPriores.</li>
<li><strong>ratios</strong> (<em>tuple of <float>, optional, default=[1]</em>) – List of aspect ratios of generated MultiBoxPriores.</li>
<li><strong>clip</strong> (<em>boolean, optional, default=0</em>) – Whether to clip out-of-boundary boxes.</li>
<li><strong>steps</strong> (<em>tuple of <float>, optional, default=[-1,-1]</em>) – Priorbox step across y and x, -1 for auto calculation.</li>
<li><strong>offsets</strong> (<em>tuple of <float>, optional, default=[0.5,0.5]</em>) – Priorbox center offsets, y and x respectively</li>
<li><strong>out</strong> (<em>NDArray, optional</em>) – The output NDArray to hold the result.</li>
</ul>
</td>
</tr>
<tr class="field-even field"><th class="field-name">Returns:</th><td class="field-body"><p class="first"><strong>out</strong> –
The output of this function.</p>
</td>
</tr>
<tr class="field-odd field"><th class="field-name">Return type:</th><td class="field-body"><p class="first last">NDArray or list of NDArrays</p>
</td>
</tr>
</tbody>
</table>
</dd></dl>
<dl class="function">
<dt id="mxnet.ndarray.contrib.MultiBoxTarget">
<code class="descclassname">mxnet.ndarray.contrib.</code><code class="descname">MultiBoxTarget</code><span class="sig-paren">(</span><em>anchor=None</em>, <em>label=None</em>, <em>cls_pred=None</em>, <em>overlap_threshold=_Null</em>, <em>ignore_label=_Null</em>, <em>negative_mining_ratio=_Null</em>, <em>negative_mining_thresh=_Null</em>, <em>minimum_negative_samples=_Null</em>, <em>variances=_Null</em>, <em>out=None</em>, <em>name=None</em>, <em>**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.ndarray.contrib.MultiBoxTarget" title="Permalink to this definition"></a></dt>
<dd><p>Compute Multibox training targets</p>
<table class="docutils field-list" frame="void" rules="none">
<col class="field-name"/>
<col class="field-body"/>
<tbody valign="top">
<tr class="field-odd field"><th class="field-name">Parameters:</th><td class="field-body"><ul class="first simple">
<li><strong>anchor</strong> (<a class="reference internal" href="ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – Generated anchor boxes.</li>
<li><strong>label</strong> (<a class="reference internal" href="ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – Object detection labels.</li>
<li><strong>cls_pred</strong> (<a class="reference internal" href="ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – Class predictions.</li>
<li><strong>overlap_threshold</strong> (<em>float, optional, default=0.5</em>) – Anchor-GT overlap threshold to be regarded as a positive match.</li>
<li><strong>ignore_label</strong> (<em>float, optional, default=-1</em>) – Label for ignored anchors.</li>
<li><strong>negative_mining_ratio</strong> (<em>float, optional, default=-1</em>) – Max negative to positive samples ratio, use -1 to disable mining</li>
<li><strong>negative_mining_thresh</strong> (<em>float, optional, default=0.5</em>) – Threshold used for negative mining.</li>
<li><strong>minimum_negative_samples</strong> (<em>int, optional, default='0'</em>) – Minimum number of negative samples.</li>
<li><strong>variances</strong> (<em>tuple of <float>, optional, default=[0.1,0.1,0.2,0.2]</em>) – Variances to be encoded in box regression target.</li>
<li><strong>out</strong> (<em>NDArray, optional</em>) – The output NDArray to hold the result.</li>
</ul>
</td>
</tr>
<tr class="field-even field"><th class="field-name">Returns:</th><td class="field-body"><p class="first"><strong>out</strong> –
The output of this function.</p>
</td>
</tr>
<tr class="field-odd field"><th class="field-name">Return type:</th><td class="field-body"><p class="first last">NDArray or list of NDArrays</p>
</td>
</tr>
</tbody>
</table>
</dd></dl>
<dl class="function">
<dt id="mxnet.ndarray.contrib.MultiProposal">
<code class="descclassname">mxnet.ndarray.contrib.</code><code class="descname">MultiProposal</code><span class="sig-paren">(</span><em>cls_score=None</em>, <em>bbox_pred=None</em>, <em>im_info=None</em>, <em>rpn_pre_nms_top_n=_Null</em>, <em>rpn_post_nms_top_n=_Null</em>, <em>threshold=_Null</em>, <em>rpn_min_size=_Null</em>, <em>scales=_Null</em>, <em>ratios=_Null</em>, <em>feature_stride=_Null</em>, <em>output_score=_Null</em>, <em>iou_loss=_Null</em>, <em>out=None</em>, <em>name=None</em>, <em>**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.ndarray.contrib.MultiProposal" title="Permalink to this definition"></a></dt>
<dd><p>Generate region proposals via RPN</p>
<table class="docutils field-list" frame="void" rules="none">
<col class="field-name"/>
<col class="field-body"/>
<tbody valign="top">
<tr class="field-odd field"><th class="field-name">Parameters:</th><td class="field-body"><ul class="first simple">
<li><strong>cls_score</strong> (<a class="reference internal" href="ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – Score of how likely proposal is object.</li>
<li><strong>bbox_pred</strong> (<a class="reference internal" href="ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – BBox Predicted deltas from anchors for proposals</li>
<li><strong>im_info</strong> (<a class="reference internal" href="ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – Image size and scale.</li>
<li><strong>rpn_pre_nms_top_n</strong> (<em>int, optional, default='6000'</em>) – Number of top scoring boxes to keep after applying NMS to RPN proposals</li>
<li><strong>rpn_post_nms_top_n</strong> (<em>int, optional, default='300'</em>) – Overlap threshold used for non-maximumsuppresion(suppress boxes with IoU >= this threshold</li>
<li><strong>threshold</strong> (<em>float, optional, default=0.7</em>) – NMS value, below which to suppress.</li>
<li><strong>rpn_min_size</strong> (<em>int, optional, default='16'</em>) – Minimum height or width in proposal</li>
<li><strong>scales</strong> (<em>tuple of <float>, optional, default=[4,8,16,32]</em>) – Used to generate anchor windows by enumerating scales</li>
<li><strong>ratios</strong> (<em>tuple of <float>, optional, default=[0.5,1,2]</em>) – Used to generate anchor windows by enumerating ratios</li>
<li><strong>feature_stride</strong> (<em>int, optional, default='16'</em>) – The size of the receptive field each unit in the convolution layer of the rpn,for example the product of all stride’s prior to this layer.</li>
<li><strong>output_score</strong> (<em>boolean, optional, default=0</em>) – Add score to outputs</li>
<li><strong>iou_loss</strong> (<em>boolean, optional, default=0</em>) – Usage of IoU Loss</li>
<li><strong>out</strong> (<em>NDArray, optional</em>) – The output NDArray to hold the result.</li>
</ul>
</td>
</tr>
<tr class="field-even field"><th class="field-name">Returns:</th><td class="field-body"><p class="first"><strong>out</strong> –
The output of this function.</p>
</td>
</tr>
<tr class="field-odd field"><th class="field-name">Return type:</th><td class="field-body"><p class="first last">NDArray or list of NDArrays</p>
</td>
</tr>
</tbody>
</table>
</dd></dl>
<dl class="function">
<dt id="mxnet.ndarray.contrib.PSROIPooling">
<code class="descclassname">mxnet.ndarray.contrib.</code><code class="descname">PSROIPooling</code><span class="sig-paren">(</span><em>data=None</em>, <em>rois=None</em>, <em>spatial_scale=_Null</em>, <em>output_dim=_Null</em>, <em>pooled_size=_Null</em>, <em>group_size=_Null</em>, <em>out=None</em>, <em>name=None</em>, <em>**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.ndarray.contrib.PSROIPooling" title="Permalink to this definition"></a></dt>
<dd><p>Performs region-of-interest pooling on inputs. Resize bounding box coordinates by spatial_scale and crop input feature maps accordingly. The cropped feature maps are pooled by max pooling to a fixed size output indicated by pooled_size. batch_size will change to the number of region bounding boxes after PSROIPooling</p>
<table class="docutils field-list" frame="void" rules="none">
<col class="field-name"/>
<col class="field-body"/>
<tbody valign="top">
<tr class="field-odd field"><th class="field-name">Parameters:</th><td class="field-body"><ul class="first simple">
<li><strong>data</strong> (<a class="reference internal" href="../symbol/symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Input data to the pooling operator, a 4D Feature maps</li>
<li><strong>rois</strong> (<a class="reference internal" href="../symbol/symbol.html#mxnet.symbol.Symbol" title="mxnet.symbol.Symbol"><em>Symbol</em></a>) – Bounding box coordinates, a 2D array of [[batch_index, x1, y1, x2, y2]]. (x1, y1) and (x2, y2) are top left and down right corners of designated region of interest. batch_index indicates the index of corresponding image in the input data</li>
<li><strong>spatial_scale</strong> (<em>float, required</em>) – Ratio of input feature map height (or w) to raw image height (or w). Equals the reciprocal of total stride in convolutional layers</li>
<li><strong>output_dim</strong> (<em>int, required</em>) – fix output dim</li>
<li><strong>pooled_size</strong> (<em>int, required</em>) – fix pooled size</li>
<li><strong>group_size</strong> (<em>int, optional, default='0'</em>) – fix group size</li>
<li><strong>out</strong> (<em>NDArray, optional</em>) – The output NDArray to hold the result.</li>
</ul>
</td>
</tr>
<tr class="field-even field"><th class="field-name">Returns:</th><td class="field-body"><p class="first"><strong>out</strong> –
The output of this function.</p>
</td>
</tr>
<tr class="field-odd field"><th class="field-name">Return type:</th><td class="field-body"><p class="first last">NDArray or list of NDArrays</p>
</td>
</tr>
</tbody>
</table>
</dd></dl>
<dl class="function">
<dt id="mxnet.ndarray.contrib.Proposal">
<code class="descclassname">mxnet.ndarray.contrib.</code><code class="descname">Proposal</code><span class="sig-paren">(</span><em>cls_score=None</em>, <em>bbox_pred=None</em>, <em>im_info=None</em>, <em>rpn_pre_nms_top_n=_Null</em>, <em>rpn_post_nms_top_n=_Null</em>, <em>threshold=_Null</em>, <em>rpn_min_size=_Null</em>, <em>scales=_Null</em>, <em>ratios=_Null</em>, <em>feature_stride=_Null</em>, <em>output_score=_Null</em>, <em>iou_loss=_Null</em>, <em>out=None</em>, <em>name=None</em>, <em>**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.ndarray.contrib.Proposal" title="Permalink to this definition"></a></dt>
<dd><p>Generate region proposals via RPN</p>
<table class="docutils field-list" frame="void" rules="none">
<col class="field-name"/>
<col class="field-body"/>
<tbody valign="top">
<tr class="field-odd field"><th class="field-name">Parameters:</th><td class="field-body"><ul class="first simple">
<li><strong>cls_score</strong> (<a class="reference internal" href="ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – Score of how likely proposal is object.</li>
<li><strong>bbox_pred</strong> (<a class="reference internal" href="ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – BBox Predicted deltas from anchors for proposals</li>
<li><strong>im_info</strong> (<a class="reference internal" href="ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – Image size and scale.</li>
<li><strong>rpn_pre_nms_top_n</strong> (<em>int, optional, default='6000'</em>) – Number of top scoring boxes to keep after applying NMS to RPN proposals</li>
<li><strong>rpn_post_nms_top_n</strong> (<em>int, optional, default='300'</em>) – Overlap threshold used for non-maximumsuppresion(suppress boxes with IoU >= this threshold</li>
<li><strong>threshold</strong> (<em>float, optional, default=0.7</em>) – NMS value, below which to suppress.</li>
<li><strong>rpn_min_size</strong> (<em>int, optional, default='16'</em>) – Minimum height or width in proposal</li>
<li><strong>scales</strong> (<em>tuple of <float>, optional, default=[4,8,16,32]</em>) – Used to generate anchor windows by enumerating scales</li>
<li><strong>ratios</strong> (<em>tuple of <float>, optional, default=[0.5,1,2]</em>) – Used to generate anchor windows by enumerating ratios</li>
<li><strong>feature_stride</strong> (<em>int, optional, default='16'</em>) – The size of the receptive field each unit in the convolution layer of the rpn,for example the product of all stride’s prior to this layer.</li>
<li><strong>output_score</strong> (<em>boolean, optional, default=0</em>) – Add score to outputs</li>
<li><strong>iou_loss</strong> (<em>boolean, optional, default=0</em>) – Usage of IoU Loss</li>
<li><strong>out</strong> (<em>NDArray, optional</em>) – The output NDArray to hold the result.</li>
</ul>
</td>
</tr>
<tr class="field-even field"><th class="field-name">Returns:</th><td class="field-body"><p class="first"><strong>out</strong> –
The output of this function.</p>
</td>
</tr>
<tr class="field-odd field"><th class="field-name">Return type:</th><td class="field-body"><p class="first last">NDArray or list of NDArrays</p>
</td>
</tr>
</tbody>
</table>
</dd></dl>
<dl class="function">
<dt id="mxnet.ndarray.contrib.SparseEmbedding">
<code class="descclassname">mxnet.ndarray.contrib.</code><code class="descname">SparseEmbedding</code><span class="sig-paren">(</span><em>data=None</em>, <em>weight=None</em>, <em>input_dim=_Null</em>, <em>output_dim=_Null</em>, <em>dtype=_Null</em>, <em>out=None</em>, <em>name=None</em>, <em>**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.ndarray.contrib.SparseEmbedding" title="Permalink to this definition"></a></dt>
<dd><p>Maps integer indices to vector representations (embeddings).</p>
<p>This operator maps words to real-valued vectors in a high-dimensional space,
called word embeddings. These embeddings can capture semantic and syntactic properties of the words.
For example, it has been noted that in the learned embedding spaces, similar words tend
to be close to each other and dissimilar words far apart.</p>
<p>For an input array of shape (d1, ..., dK),
the shape of an output array is (d1, ..., dK, output_dim).
All the input values should be integers in the range [0, input_dim).</p>
<p>If the input_dim is ip0 and output_dim is op0, then shape of the embedding weight matrix must be
(ip0, op0).</p>
<p>The storage type of weight must be <cite>row_sparse</cite>, and the gradient of the weight will be of
<cite>row_sparse</cite> storage type, too.</p>
<div class="admonition note">
<p class="first admonition-title">Note</p>
<p class="last"><cite>SparseEmbedding</cite> is designed for the use case where <cite>input_dim</cite> is very large (e.g. 100k).
The operator is available on both CPU and GPU.</p>
</div>
<p>Examples:</p>
<div class="highlight-python"><div class="highlight"><pre><span></span>input_dim = 4
output_dim = 5
// Each row in weight matrix y represents a word. So, y = (w0,w1,w2,w3)
y = [[ 0., 1., 2., 3., 4.],
[ 5., 6., 7., 8., 9.],
[ 10., 11., 12., 13., 14.],
[ 15., 16., 17., 18., 19.]]
// Input array x represents n-grams(2-gram). So, x = [(w1,w3), (w0,w2)]
x = [[ 1., 3.],
[ 0., 2.]]
// Mapped input x to its vector representation y.
SparseEmbedding(x, y, 4, 5) = [[[ 5., 6., 7., 8., 9.],
[ 15., 16., 17., 18., 19.]],
[[ 0., 1., 2., 3., 4.],
[ 10., 11., 12., 13., 14.]]]
</pre></div>
</div>
<p>Defined in src/operator/tensor/indexing_op.cc:L294</p>
<table class="docutils field-list" frame="void" rules="none">
<col class="field-name"/>
<col class="field-body"/>
<tbody valign="top">
<tr class="field-odd field"><th class="field-name">Parameters:</th><td class="field-body"><ul class="first simple">
<li><strong>data</strong> (<a class="reference internal" href="ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – The input array to the embedding operator.</li>
<li><strong>weight</strong> (<a class="reference internal" href="ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – The embedding weight matrix.</li>
<li><strong>input_dim</strong> (<em>int, required</em>) – Vocabulary size of the input indices.</li>
<li><strong>output_dim</strong> (<em>int, required</em>) – Dimension of the embedding vectors.</li>
<li><strong>dtype</strong> (<em>{'float16', 'float32', 'float64', 'int32', 'int64', 'int8', 'uint8'},optional, default='float32'</em>) – Data type of weight.</li>
<li><strong>out</strong> (<em>NDArray, optional</em>) – The output NDArray to hold the result.</li>
</ul>
</td>
</tr>
<tr class="field-even field"><th class="field-name">Returns:</th><td class="field-body"><p class="first"><strong>out</strong> –
The output of this function.</p>
</td>
</tr>
<tr class="field-odd field"><th class="field-name">Return type:</th><td class="field-body"><p class="first last">NDArray or list of NDArrays</p>
</td>
</tr>
</tbody>
</table>
</dd></dl>
<dl class="function">
<dt id="mxnet.ndarray.contrib.bipartite_matching">
<code class="descclassname">mxnet.ndarray.contrib.</code><code class="descname">bipartite_matching</code><span class="sig-paren">(</span><em>data=None</em>, <em>is_ascend=_Null</em>, <em>threshold=_Null</em>, <em>topk=_Null</em>, <em>out=None</em>, <em>name=None</em>, <em>**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.ndarray.contrib.bipartite_matching" title="Permalink to this definition"></a></dt>
<dd><dl class="docutils">
<dt>Compute bipartite matching.</dt>
<dd><p class="first">The matching is performed on score matrix with shape [B, N, M]
- B: batch_size
- N: number of rows to match
- M: number of columns as reference to be matched against.</p>
<p>Returns:
x : matched column indices. -1 indicating non-matched elements in rows.
y : matched row indices.</p>
<p>Note:</p>
<div class="highlight-python"><div class="highlight"><pre><span></span>Zero gradients are back-propagated in this op for now.
</pre></div>
</div>
<p>Example:</p>
<div class="last highlight-python"><div class="highlight"><pre><span></span><span class="n">s</span> <span class="o">=</span> <span class="p">[[</span><span class="mf">0.5</span><span class="p">,</span> <span class="mf">0.6</span><span class="p">],</span> <span class="p">[</span><span class="mf">0.1</span><span class="p">,</span> <span class="mf">0.2</span><span class="p">],</span> <span class="p">[</span><span class="mf">0.3</span><span class="p">,</span> <span class="mf">0.4</span><span class="p">]]</span>
<span class="n">x</span><span class="p">,</span> <span class="n">y</span> <span class="o">=</span> <span class="n">bipartite_matching</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">threshold</span><span class="o">=</span><span class="mf">1e-12</span><span class="p">,</span> <span class="n">is_ascend</span><span class="o">=</span><span class="bp">False</span><span class="p">)</span>
<span class="n">x</span> <span class="o">=</span> <span class="p">[</span><span class="mi">1</span><span class="p">,</span> <span class="o">-</span><span class="mi">1</span><span class="p">,</span> <span class="mi">0</span><span class="p">]</span>
<span class="n">y</span> <span class="o">=</span> <span class="p">[</span><span class="mi">2</span><span class="p">,</span> <span class="mi">0</span><span class="p">]</span>
</pre></div>
</div>
</dd>
</dl>
<p>Defined in src/operator/contrib/bounding_box.cc:L169</p>
<table class="docutils field-list" frame="void" rules="none">
<col class="field-name"/>
<col class="field-body"/>
<tbody valign="top">
<tr class="field-odd field"><th class="field-name">Parameters:</th><td class="field-body"><ul class="first simple">
<li><strong>data</strong> (<a class="reference internal" href="ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – The input</li>
<li><strong>is_ascend</strong> (<em>boolean, optional, default=0</em>) – Use ascend order for scores instead of descending. Please set threshold accordingly.</li>
<li><strong>threshold</strong> (<em>float, required</em>) – Ignore matching when score < thresh, if is_ascend=false, or ignore score > thresh, if is_ascend=true.</li>
<li><strong>topk</strong> (<em>int, optional, default='-1'</em>) – Limit the number of matches to topk, set -1 for no limit</li>
<li><strong>out</strong> (<em>NDArray, optional</em>) – The output NDArray to hold the result.</li>
</ul>
</td>
</tr>
<tr class="field-even field"><th class="field-name">Returns:</th><td class="field-body"><p class="first"><strong>out</strong> –
The output of this function.</p>
</td>
</tr>
<tr class="field-odd field"><th class="field-name">Return type:</th><td class="field-body"><p class="first last">NDArray or list of NDArrays</p>
</td>
</tr>
</tbody>
</table>
</dd></dl>
<dl class="function">
<dt id="mxnet.ndarray.contrib.box_iou">
<code class="descclassname">mxnet.ndarray.contrib.</code><code class="descname">box_iou</code><span class="sig-paren">(</span><em>lhs=None</em>, <em>rhs=None</em>, <em>format=_Null</em>, <em>out=None</em>, <em>name=None</em>, <em>**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.ndarray.contrib.box_iou" title="Permalink to this definition"></a></dt>
<dd><dl class="docutils">
<dt>Bounding box overlap of two arrays.</dt>
<dd><p class="first">The overlap is defined as Intersection-over-Union, aka, IOU.
- lhs: (a_1, a_2, ..., a_n, 4) array
- rhs: (b_1, b_2, ..., b_n, 4) array
- output: (a_1, a_2, ..., a_n, b_1, b_2, ..., b_n) array</p>
<p>Note:</p>
<div class="highlight-python"><div class="highlight"><pre><span></span>Zero gradients are back-propagated in this op for now.
</pre></div>
</div>
<p>Example:</p>
<div class="last highlight-python"><div class="highlight"><pre><span></span><span class="n">x</span> <span class="o">=</span> <span class="p">[[</span><span class="mf">0.5</span><span class="p">,</span> <span class="mf">0.5</span><span class="p">,</span> <span class="mf">1.0</span><span class="p">,</span> <span class="mf">1.0</span><span class="p">],</span> <span class="p">[</span><span class="mf">0.0</span><span class="p">,</span> <span class="mf">0.0</span><span class="p">,</span> <span class="mf">0.5</span><span class="p">,</span> <span class="mf">0.5</span><span class="p">]]</span>
<span class="n">y</span> <span class="o">=</span> <span class="p">[</span><span class="mf">0.25</span><span class="p">,</span> <span class="mf">0.25</span><span class="p">,</span> <span class="mf">0.75</span><span class="p">,</span> <span class="mf">0.75</span><span class="p">]</span>
<span class="n">box_iou</span><span class="p">(</span><span class="n">x</span><span class="p">,</span> <span class="n">y</span><span class="p">,</span> <span class="n">format</span><span class="o">=</span><span class="s1">'corner'</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span><span class="mf">0.1428</span><span class="p">],</span> <span class="p">[</span><span class="mf">0.1428</span><span class="p">]]</span>
</pre></div>
</div>
</dd>
</dl>
<p>Defined in src/operator/contrib/bounding_box.cc:L123</p>
<table class="docutils field-list" frame="void" rules="none">
<col class="field-name"/>
<col class="field-body"/>
<tbody valign="top">
<tr class="field-odd field"><th class="field-name">Parameters:</th><td class="field-body"><ul class="first simple">
<li><strong>lhs</strong> (<a class="reference internal" href="ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – The first input</li>
<li><strong>rhs</strong> (<a class="reference internal" href="ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – The second input</li>
<li><strong>format</strong> (<em>{'center', 'corner'},optional, default='corner'</em>) – The box encoding type.
“corner” means boxes are encoded as [xmin, ymin, xmax, ymax], “center” means boxes are encodes as [x, y, width, height].</li>
<li><strong>out</strong> (<em>NDArray, optional</em>) – The output NDArray to hold the result.</li>
</ul>
</td>
</tr>
<tr class="field-even field"><th class="field-name">Returns:</th><td class="field-body"><p class="first"><strong>out</strong> –
The output of this function.</p>
</td>
</tr>
<tr class="field-odd field"><th class="field-name">Return type:</th><td class="field-body"><p class="first last">NDArray or list of NDArrays</p>
</td>
</tr>
</tbody>
</table>
</dd></dl>
<dl class="function">
<dt id="mxnet.ndarray.contrib.box_nms">
<code class="descclassname">mxnet.ndarray.contrib.</code><code class="descname">box_nms</code><span class="sig-paren">(</span><em>data=None</em>, <em>overlap_thresh=_Null</em>, <em>topk=_Null</em>, <em>coord_start=_Null</em>, <em>score_index=_Null</em>, <em>id_index=_Null</em>, <em>force_suppress=_Null</em>, <em>in_format=_Null</em>, <em>out_format=_Null</em>, <em>out=None</em>, <em>name=None</em>, <em>**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.ndarray.contrib.box_nms" title="Permalink to this definition"></a></dt>
<dd><p>Apply non-maximum suppression to input.</p>
<p>The output will be sorted in descending order according to <cite>score</cite>. Boxes with
overlaps larger than <cite>overlap_thresh</cite> and smaller scores will be removed and
filled with -1, the corresponding position will be recorded for backward propogation.</p>
<p>During back-propagation, the gradient will be copied to the original
position according to the input index. For positions that have been suppressed,
the in_grad will be assigned 0.
In summary, gradients are sticked to its boxes, will either be moved or discarded
according to its original index in input.</p>
<p>Input requirements:
1. Input tensor have at least 2 dimensions, (n, k), any higher dims will be regarded
as batch, e.g. (a, b, c, d, n, k) == (a*b*c*d, n, k)
2. n is the number of boxes in each batch
3. k is the width of each box item.</p>
<p>By default, a box is [id, score, xmin, ymin, xmax, ymax, ...],
additional elements are allowed.
- <cite>id_index</cite>: optional, use -1 to ignore, useful if <cite>force_suppress=False</cite>, which means
we will skip highly overlapped boxes if one is <cite>apple</cite> while the other is <cite>car</cite>.
- <cite>coord_start</cite>: required, default=2, the starting index of the 4 coordinates.
Two formats are supported:</p>
<blockquote>
<div><cite>corner</cite>: [xmin, ymin, xmax, ymax]
<cite>center</cite>: [x, y, width, height]</div></blockquote>
<ul class="simple">
<li><cite>score_index</cite>: required, default=1, box score/confidence.</li>
</ul>
<p>When two boxes overlap IOU > <cite>overlap_thresh</cite>, the one with smaller score will be suppressed.
- <cite>in_format</cite> and <cite>out_format</cite>: default=’corner’, specify in/out box formats.</p>
<p>Examples:</p>
<div class="highlight-python"><div class="highlight"><pre><span></span>x = [[0, 0.5, 0.1, 0.1, 0.2, 0.2], [1, 0.4, 0.1, 0.1, 0.2, 0.2],
[0, 0.3, 0.1, 0.1, 0.14, 0.14], [2, 0.6, 0.5, 0.5, 0.7, 0.8]]
box_nms(x, overlap_thresh=0.1, coord_start=2, score_index=1, id_index=0,
force_suppress=True, in_format='corner', out_typ='corner') =
[[2, 0.6, 0.5, 0.5, 0.7, 0.8], [0, 0.5, 0.1, 0.1, 0.2, 0.2],
[-1, -1, -1, -1, -1, -1], [-1, -1, -1, -1, -1, -1]]
out_grad = [[0.1, 0.1, 0.1, 0.1, 0.1, 0.1], [0.2, 0.2, 0.2, 0.2, 0.2, 0.2],
[0.3, 0.3, 0.3, 0.3, 0.3, 0.3], [0.4, 0.4, 0.4, 0.4, 0.4, 0.4]]
# exe.backward
in_grad = [[0.2, 0.2, 0.2, 0.2, 0.2, 0.2], [0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 0, 0], [0.1, 0.1, 0.1, 0.1, 0.1, 0.1]]
</pre></div>
</div>
<p>Defined in src/operator/contrib/bounding_box.cc:L82</p>
<table class="docutils field-list" frame="void" rules="none">
<col class="field-name"/>
<col class="field-body"/>
<tbody valign="top">
<tr class="field-odd field"><th class="field-name">Parameters:</th><td class="field-body"><ul class="first simple">
<li><strong>data</strong> (<a class="reference internal" href="ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – The input</li>
<li><strong>overlap_thresh</strong> (<em>float, optional, default=0.5</em>) – Overlapping(IoU) threshold to suppress object with smaller score.</li>
<li><strong>topk</strong> (<em>int, optional, default='-1'</em>) – Apply nms to topk boxes with descending scores, -1 to no restriction.</li>
<li><strong>coord_start</strong> (<em>int, optional, default='2'</em>) – Start index of the consecutive 4 coordinates.</li>
<li><strong>score_index</strong> (<em>int, optional, default='1'</em>) – Index of the scores/confidence of boxes.</li>
<li><strong>id_index</strong> (<em>int, optional, default='-1'</em>) – Optional, index of the class categories, -1 to disable.</li>
<li><strong>force_suppress</strong> (<em>boolean, optional, default=0</em>) – Optional, if set false and id_index is provided, nms will only apply to boxes belongs to the same category</li>
<li><strong>in_format</strong> (<em>{'center', 'corner'},optional, default='corner'</em>) – The input box encoding type.
“corner” means boxes are encoded as [xmin, ymin, xmax, ymax], “center” means boxes are encodes as [x, y, width, height].</li>
<li><strong>out_format</strong> (<em>{'center', 'corner'},optional, default='corner'</em>) – The output box encoding type.
“corner” means boxes are encoded as [xmin, ymin, xmax, ymax], “center” means boxes are encodes as [x, y, width, height].</li>
<li><strong>out</strong> (<em>NDArray, optional</em>) – The output NDArray to hold the result.</li>
</ul>
</td>
</tr>
<tr class="field-even field"><th class="field-name">Returns:</th><td class="field-body"><p class="first"><strong>out</strong> –
The output of this function.</p>
</td>
</tr>
<tr class="field-odd field"><th class="field-name">Return type:</th><td class="field-body"><p class="first last">NDArray or list of NDArrays</p>
</td>
</tr>
</tbody>
</table>
</dd></dl>
<dl class="function">
<dt id="mxnet.ndarray.contrib.box_non_maximum_suppression">
<code class="descclassname">mxnet.ndarray.contrib.</code><code class="descname">box_non_maximum_suppression</code><span class="sig-paren">(</span><em>data=None</em>, <em>overlap_thresh=_Null</em>, <em>topk=_Null</em>, <em>coord_start=_Null</em>, <em>score_index=_Null</em>, <em>id_index=_Null</em>, <em>force_suppress=_Null</em>, <em>in_format=_Null</em>, <em>out_format=_Null</em>, <em>out=None</em>, <em>name=None</em>, <em>**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.ndarray.contrib.box_non_maximum_suppression" title="Permalink to this definition"></a></dt>
<dd><p>Apply non-maximum suppression to input.</p>
<p>The output will be sorted in descending order according to <cite>score</cite>. Boxes with
overlaps larger than <cite>overlap_thresh</cite> and smaller scores will be removed and
filled with -1, the corresponding position will be recorded for backward propogation.</p>
<p>During back-propagation, the gradient will be copied to the original
position according to the input index. For positions that have been suppressed,
the in_grad will be assigned 0.
In summary, gradients are sticked to its boxes, will either be moved or discarded
according to its original index in input.</p>
<p>Input requirements:
1. Input tensor have at least 2 dimensions, (n, k), any higher dims will be regarded
as batch, e.g. (a, b, c, d, n, k) == (a*b*c*d, n, k)
2. n is the number of boxes in each batch
3. k is the width of each box item.</p>
<p>By default, a box is [id, score, xmin, ymin, xmax, ymax, ...],
additional elements are allowed.
- <cite>id_index</cite>: optional, use -1 to ignore, useful if <cite>force_suppress=False</cite>, which means
we will skip highly overlapped boxes if one is <cite>apple</cite> while the other is <cite>car</cite>.
- <cite>coord_start</cite>: required, default=2, the starting index of the 4 coordinates.
Two formats are supported:</p>
<blockquote>
<div><cite>corner</cite>: [xmin, ymin, xmax, ymax]
<cite>center</cite>: [x, y, width, height]</div></blockquote>
<ul class="simple">
<li><cite>score_index</cite>: required, default=1, box score/confidence.</li>
</ul>
<p>When two boxes overlap IOU > <cite>overlap_thresh</cite>, the one with smaller score will be suppressed.
- <cite>in_format</cite> and <cite>out_format</cite>: default=’corner’, specify in/out box formats.</p>
<p>Examples:</p>
<div class="highlight-python"><div class="highlight"><pre><span></span>x = [[0, 0.5, 0.1, 0.1, 0.2, 0.2], [1, 0.4, 0.1, 0.1, 0.2, 0.2],
[0, 0.3, 0.1, 0.1, 0.14, 0.14], [2, 0.6, 0.5, 0.5, 0.7, 0.8]]
box_nms(x, overlap_thresh=0.1, coord_start=2, score_index=1, id_index=0,
force_suppress=True, in_format='corner', out_typ='corner') =
[[2, 0.6, 0.5, 0.5, 0.7, 0.8], [0, 0.5, 0.1, 0.1, 0.2, 0.2],
[-1, -1, -1, -1, -1, -1], [-1, -1, -1, -1, -1, -1]]
out_grad = [[0.1, 0.1, 0.1, 0.1, 0.1, 0.1], [0.2, 0.2, 0.2, 0.2, 0.2, 0.2],
[0.3, 0.3, 0.3, 0.3, 0.3, 0.3], [0.4, 0.4, 0.4, 0.4, 0.4, 0.4]]
# exe.backward
in_grad = [[0.2, 0.2, 0.2, 0.2, 0.2, 0.2], [0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 0, 0], [0.1, 0.1, 0.1, 0.1, 0.1, 0.1]]
</pre></div>
</div>
<p>Defined in src/operator/contrib/bounding_box.cc:L82</p>
<table class="docutils field-list" frame="void" rules="none">
<col class="field-name"/>
<col class="field-body"/>
<tbody valign="top">
<tr class="field-odd field"><th class="field-name">Parameters:</th><td class="field-body"><ul class="first simple">
<li><strong>data</strong> (<a class="reference internal" href="ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – The input</li>
<li><strong>overlap_thresh</strong> (<em>float, optional, default=0.5</em>) – Overlapping(IoU) threshold to suppress object with smaller score.</li>
<li><strong>topk</strong> (<em>int, optional, default='-1'</em>) – Apply nms to topk boxes with descending scores, -1 to no restriction.</li>
<li><strong>coord_start</strong> (<em>int, optional, default='2'</em>) – Start index of the consecutive 4 coordinates.</li>
<li><strong>score_index</strong> (<em>int, optional, default='1'</em>) – Index of the scores/confidence of boxes.</li>
<li><strong>id_index</strong> (<em>int, optional, default='-1'</em>) – Optional, index of the class categories, -1 to disable.</li>
<li><strong>force_suppress</strong> (<em>boolean, optional, default=0</em>) – Optional, if set false and id_index is provided, nms will only apply to boxes belongs to the same category</li>
<li><strong>in_format</strong> (<em>{'center', 'corner'},optional, default='corner'</em>) – The input box encoding type.
“corner” means boxes are encoded as [xmin, ymin, xmax, ymax], “center” means boxes are encodes as [x, y, width, height].</li>
<li><strong>out_format</strong> (<em>{'center', 'corner'},optional, default='corner'</em>) – The output box encoding type.
“corner” means boxes are encoded as [xmin, ymin, xmax, ymax], “center” means boxes are encodes as [x, y, width, height].</li>
<li><strong>out</strong> (<em>NDArray, optional</em>) – The output NDArray to hold the result.</li>
</ul>
</td>
</tr>
<tr class="field-even field"><th class="field-name">Returns:</th><td class="field-body"><p class="first"><strong>out</strong> –
The output of this function.</p>
</td>
</tr>
<tr class="field-odd field"><th class="field-name">Return type:</th><td class="field-body"><p class="first last">NDArray or list of NDArrays</p>
</td>
</tr>
</tbody>
</table>
</dd></dl>
<dl class="function">
<dt id="mxnet.ndarray.contrib.count_sketch">
<code class="descclassname">mxnet.ndarray.contrib.</code><code class="descname">count_sketch</code><span class="sig-paren">(</span><em>data=None</em>, <em>h=None</em>, <em>s=None</em>, <em>out_dim=_Null</em>, <em>processing_batch_size=_Null</em>, <em>out=None</em>, <em>name=None</em>, <em>**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.ndarray.contrib.count_sketch" title="Permalink to this definition"></a></dt>
<dd><p>Apply CountSketch to input: map a d-dimension data to k-dimension data”</p>
<div class="admonition note">
<p class="first admonition-title">Note</p>
<p class="last"><cite>count_sketch</cite> is only available on GPU.</p>
</div>
<p>Assume input data has shape (N, d), sign hash table s has shape (N, d),
index hash table h has shape (N, d) and mapping dimension out_dim = k,
each element in s is either +1 or -1, each element in h is random integer from 0 to k-1.
Then the operator computs:</p>
<div class="math">
\[out[h[i]] += data[i] * s[i]\]</div>
<p>Example:</p>
<div class="highlight-python"><div class="highlight"><pre><span></span><span class="n">out_dim</span> <span class="o">=</span> <span class="mi">5</span>
<span class="n">x</span> <span class="o">=</span> <span class="p">[[</span><span class="mf">1.2</span><span class="p">,</span> <span class="mf">2.5</span><span class="p">,</span> <span class="mf">3.4</span><span class="p">],[</span><span class="mf">3.2</span><span class="p">,</span> <span class="mf">5.7</span><span class="p">,</span> <span class="mf">6.6</span><span class="p">]]</span>
<span class="n">h</span> <span class="o">=</span> <span class="p">[[</span><span class="mi">0</span><span class="p">,</span> <span class="mi">3</span><span class="p">,</span> <span class="mi">4</span><span class="p">]]</span>
<span class="n">s</span> <span class="o">=</span> <span class="p">[[</span><span class="mi">1</span><span class="p">,</span> <span class="o">-</span><span class="mi">1</span><span class="p">,</span> <span class="mi">1</span><span class="p">]]</span>
<span class="n">mx</span><span class="o">.</span><span class="n">contrib</span><span class="o">.</span><span class="n">ndarray</span><span class="o">.</span><span class="n">count_sketch</span><span class="p">(</span><span class="n">data</span><span class="o">=</span><span class="n">x</span><span class="p">,</span> <span class="n">h</span><span class="o">=</span><span class="n">h</span><span class="p">,</span> <span class="n">s</span><span class="o">=</span><span class="n">s</span><span class="p">,</span> <span class="n">out_dim</span> <span class="o">=</span> <span class="mi">5</span><span class="p">)</span> <span class="o">=</span> <span class="p">[[</span><span class="mf">1.2</span><span class="p">,</span> <span class="mi">0</span><span class="p">,</span> <span class="mi">0</span><span class="p">,</span> <span class="o">-</span><span class="mf">2.5</span><span class="p">,</span> <span class="mf">3.4</span><span class="p">],</span>
<span class="p">[</span><span class="mf">3.2</span><span class="p">,</span> <span class="mi">0</span><span class="p">,</span> <span class="mi">0</span><span class="p">,</span> <span class="o">-</span><span class="mf">5.7</span><span class="p">,</span> <span class="mf">6.6</span><span class="p">]]</span>
</pre></div>
</div>
<p>Defined in src/operator/contrib/count_sketch.cc:L67</p>
<table class="docutils field-list" frame="void" rules="none">
<col class="field-name"/>
<col class="field-body"/>
<tbody valign="top">
<tr class="field-odd field"><th class="field-name">Parameters:</th><td class="field-body"><ul class="first simple">
<li><strong>data</strong> (<a class="reference internal" href="ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – Input data to the CountSketchOp.</li>
<li><strong>h</strong> (<a class="reference internal" href="ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – The index vector</li>
<li><strong>s</strong> (<a class="reference internal" href="ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – The sign vector</li>
<li><strong>out_dim</strong> (<em>int, required</em>) – The output dimension.</li>
<li><strong>processing_batch_size</strong> (<em>int, optional, default='32'</em>) – How many sketch vectors to process at one time.</li>
<li><strong>out</strong> (<em>NDArray, optional</em>) – The output NDArray to hold the result.</li>
</ul>
</td>
</tr>
<tr class="field-even field"><th class="field-name">Returns:</th><td class="field-body"><p class="first"><strong>out</strong> –
The output of this function.</p>
</td>
</tr>
<tr class="field-odd field"><th class="field-name">Return type:</th><td class="field-body"><p class="first last">NDArray or list of NDArrays</p>
</td>
</tr>
</tbody>
</table>
</dd></dl>
<dl class="function">
<dt id="mxnet.ndarray.contrib.ctc_loss">
<code class="descclassname">mxnet.ndarray.contrib.</code><code class="descname">ctc_loss</code><span class="sig-paren">(</span><em>data=None</em>, <em>label=None</em>, <em>data_lengths=None</em>, <em>label_lengths=None</em>, <em>use_data_lengths=_Null</em>, <em>use_label_lengths=_Null</em>, <em>blank_label=_Null</em>, <em>out=None</em>, <em>name=None</em>, <em>**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.ndarray.contrib.ctc_loss" title="Permalink to this definition"></a></dt>
<dd><p>Connectionist Temporal Classification Loss.</p>
<p>The shapes of the inputs and outputs:</p>
<ul class="simple">
<li><strong>data</strong>: <cite>(sequence_length, batch_size, alphabet_size)</cite></li>
<li><strong>label</strong>: <cite>(batch_size, label_sequence_length)</cite></li>
<li><strong>out</strong>: <cite>(batch_size)</cite></li>
</ul>
<p>The <cite>data</cite> tensor consists of sequences of activation vectors (without applying softmax),
with i-th channel in the last dimension corresponding to i-th label
for i between 0 and alphabet_size-1 (i.e always 0-indexed).
Alphabet size should include one additional value reserved for blank label.
When <cite>blank_label</cite> is <code class="docutils literal"><span class="pre">"first"</span></code>, the <code class="docutils literal"><span class="pre">0</span></code>-th channel is be reserved for
activation of blank label, or otherwise if it is “last”, <code class="docutils literal"><span class="pre">(alphabet_size-1)</span></code>-th channel should be
reserved for blank label.</p>
<p><code class="docutils literal"><span class="pre">label</span></code> is an index matrix of integers. When <cite>blank_label</cite> is <code class="docutils literal"><span class="pre">"first"</span></code>,
the value 0 is then reserved for blank label, and should not be passed in this matrix. Otherwise,
when <cite>blank_label</cite> is <code class="docutils literal"><span class="pre">"last"</span></code>, the value <cite>(alphabet_size-1)</cite> is reserved for blank label.</p>
<p>If a sequence of labels is shorter than <em>label_sequence_length</em>, use the special
padding value at the end of the sequence to conform it to the correct
length. The padding value is <cite>0</cite> when <cite>blank_label</cite> is <code class="docutils literal"><span class="pre">"first"</span></code>, and <cite>-1</cite> otherwise.</p>
<p>For example, suppose the vocabulary is <cite>[a, b, c]</cite>, and in one batch we have three sequences
‘ba’, ‘cbb’, and ‘abac’. When <cite>blank_label</cite> is <code class="docutils literal"><span class="pre">"first"</span></code>, we can index the labels as
<cite>{‘a’: 1, ‘b’: 2, ‘c’: 3}</cite>, and we reserve the 0-th channel for blank label in data tensor.
The resulting <cite>label</cite> tensor should be padded to be:</p>
<div class="highlight-python"><div class="highlight"><pre><span></span><span class="p">[[</span><span class="mi">2</span><span class="p">,</span> <span class="mi">1</span><span class="p">,</span> <span class="mi">0</span><span class="p">,</span> <span class="mi">0</span><span class="p">],</span> <span class="p">[</span><span class="mi">3</span><span class="p">,</span> <span class="mi">2</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="p">[</span><span class="mi">1</span><span class="p">,</span> <span class="mi">2</span><span class="p">,</span> <span class="mi">1</span><span class="p">,</span> <span class="mi">3</span><span class="p">]]</span>
</pre></div>
</div>
<p>When <cite>blank_label</cite> is <code class="docutils literal"><span class="pre">"last"</span></code>, we can index the labels as
<cite>{‘a’: 0, ‘b’: 1, ‘c’: 2}</cite>, and we reserve the channel index 3 for blank label in data tensor.
The resulting <cite>label</cite> tensor should be padded to be:</p>
<div class="highlight-python"><div class="highlight"><pre><span></span><span class="p">[[</span><span class="mi">1</span><span class="p">,</span> <span class="mi">0</span><span class="p">,</span> <span class="o">-</span><span class="mi">1</span><span class="p">,</span> <span class="o">-</span><span class="mi">1</span><span class="p">],</span> <span class="p">[</span><span class="mi">2</span><span class="p">,</span> <span class="mi">1</span><span class="p">,</span> <span class="mi">1</span><span class="p">,</span> <span class="o">-</span><span class="mi">1</span><span class="p">],</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="mi">0</span><span class="p">,</span> <span class="mi">2</span><span class="p">]]</span>
</pre></div>
</div>
<p><code class="docutils literal"><span class="pre">out</span></code> is a list of CTC loss values, one per example in the batch.</p>
<p>See <em>Connectionist Temporal Classification: Labelling Unsegmented
Sequence Data with Recurrent Neural Networks</em>, A. Graves <em>et al</em>. for more
information on the definition and the algorithm.</p>
<p>Defined in src/operator/contrib/ctc_loss.cc:L115</p>
<table class="docutils field-list" frame="void" rules="none">
<col class="field-name"/>
<col class="field-body"/>
<tbody valign="top">
<tr class="field-odd field"><th class="field-name">Parameters:</th><td class="field-body"><ul class="first simple">
<li><strong>data</strong> (<a class="reference internal" href="ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – Input data to the ctc_loss op.</li>
<li><strong>label</strong> (<a class="reference internal" href="ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – Ground-truth labels for the loss.</li>
<li><strong>data_lengths</strong> (<a class="reference internal" href="ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – Lengths of data for each of the samples. Only required when use_data_lengths is true.</li>
<li><strong>label_lengths</strong> (<a class="reference internal" href="ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – Lengths of labels for each of the samples. Only required when use_label_lengths is true.</li>
<li><strong>use_data_lengths</strong> (<em>boolean, optional, default=0</em>) – Whether the data lenghts are decided by <cite>data_lengths</cite>. If false, the lengths are equal to the max sequence length.</li>
<li><strong>use_label_lengths</strong> (<em>boolean, optional, default=0</em>) – Whether the label lenghts are decided by <cite>label_lengths</cite>, or derived from <cite>padding_mask</cite>. If false, the lengths are derived from the first occurrence of the value of <cite>padding_mask</cite>. The value of <cite>padding_mask</cite> is <code class="docutils literal"><span class="pre">0</span></code> when first CTC label is reserved for blank, and <code class="docutils literal"><span class="pre">-1</span></code> when last label is reserved for blank. See <cite>blank_label</cite>.</li>
<li><strong>blank_label</strong> (<em>{'first', 'last'},optional, default='first'</em>) – Set the label that is reserved for blank label.If “first”, 0-th label is reserved, and label values for tokens in the vocabulary are between <code class="docutils literal"><span class="pre">1</span></code> and <code class="docutils literal"><span class="pre">alphabet_size-1</span></code>, and the padding mask is <code class="docutils literal"><span class="pre">-1</span></code>. If “last”, last label value <code class="docutils literal"><span class="pre">alphabet_size-1</span></code> is reserved for blank label instead, and label values for tokens in the vocabulary are between <code class="docutils literal"><span class="pre">0</span></code> and <code class="docutils literal"><span class="pre">alphabet_size-2</span></code>, and the padding mask is <code class="docutils literal"><span class="pre">0</span></code>.</li>
<li><strong>out</strong> (<em>NDArray, optional</em>) – The output NDArray to hold the result.</li>
</ul>
</td>
</tr>
<tr class="field-even field"><th class="field-name">Returns:</th><td class="field-body"><p class="first"><strong>out</strong> –
The output of this function.</p>
</td>
</tr>
<tr class="field-odd field"><th class="field-name">Return type:</th><td class="field-body"><p class="first last">NDArray or list of NDArrays</p>
</td>
</tr>
</tbody>
</table>
</dd></dl>
<dl class="function">
<dt id="mxnet.ndarray.contrib.dequantize">
<code class="descclassname">mxnet.ndarray.contrib.</code><code class="descname">dequantize</code><span class="sig-paren">(</span><em>input=None</em>, <em>min_range=None</em>, <em>max_range=None</em>, <em>out_type=_Null</em>, <em>out=None</em>, <em>name=None</em>, <em>**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.ndarray.contrib.dequantize" title="Permalink to this definition"></a></dt>
<dd><p>Dequantize the input tensor into a float tensor.
[min_range, max_range] are scalar floats that spcify the range for
the output data.</p>
<p>Each value of the tensor will undergo the following:</p>
<p><cite>out[i] = min_range + (in[i] * (max_range - min_range) / range(INPUT_TYPE))</cite></p>
<p>here <cite>range(T) = numeric_limits<T>::max() - numeric_limits<T>::min()</cite></p>
<p>Defined in src/operator/contrib/dequantize.cc:L41</p>
<table class="docutils field-list" frame="void" rules="none">
<col class="field-name"/>
<col class="field-body"/>
<tbody valign="top">
<tr class="field-odd field"><th class="field-name">Parameters:</th><td class="field-body"><ul class="first simple">
<li><strong>input</strong> (<a class="reference internal" href="ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – A ndarray/symbol of type <cite>uint8</cite></li>
<li><strong>min_range</strong> (<a class="reference internal" href="ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – The minimum scalar value possibly produced for the input</li>
<li><strong>max_range</strong> (<a class="reference internal" href="ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – The maximum scalar value possibly produced for the input</li>
<li><strong>out_type</strong> (<em>{'float32'}, required</em>) – Output data type.</li>
<li><strong>out</strong> (<em>NDArray, optional</em>) – The output NDArray to hold the result.</li>
</ul>
</td>
</tr>
<tr class="field-even field"><th class="field-name">Returns:</th><td class="field-body"><p class="first"><strong>out</strong> –
The output of this function.</p>
</td>
</tr>
<tr class="field-odd field"><th class="field-name">Return type:</th><td class="field-body"><p class="first last">NDArray or list of NDArrays</p>
</td>
</tr>
</tbody>
</table>
</dd></dl>
<dl class="function">
<dt id="mxnet.ndarray.contrib.fft">
<code class="descclassname">mxnet.ndarray.contrib.</code><code class="descname">fft</code><span class="sig-paren">(</span><em>data=None</em>, <em>compute_size=_Null</em>, <em>out=None</em>, <em>name=None</em>, <em>**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.ndarray.contrib.fft" title="Permalink to this definition"></a></dt>
<dd><p>Apply 1D FFT to input”</p>
<div class="admonition note">
<p class="first admonition-title">Note</p>
<p class="last"><cite>fft</cite> is only available on GPU.</p>
</div>
<p>Currently accept 2 input data shapes: (N, d) or (N1, N2, N3, d), data can only be real numbers.
The output data has shape: (N, 2*d) or (N1, N2, N3, 2*d). The format is: [real0, imag0, real1, imag1, ...].</p>
<p>Example:</p>
<div class="highlight-python"><div class="highlight"><pre><span></span><span class="n">data</span> <span class="o">=</span> <span class="n">np</span><span class="o">.</span><span class="n">random</span><span class="o">.</span><span class="n">normal</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="mi">3</span><span class="p">,</span><span class="mi">4</span><span class="p">))</span>
<span class="n">out</span> <span class="o">=</span> <span class="n">mx</span><span class="o">.</span><span class="n">contrib</span><span class="o">.</span><span class="n">ndarray</span><span class="o">.</span><span class="n">fft</span><span class="p">(</span><span class="n">data</span> <span class="o">=</span> <span class="n">mx</span><span class="o">.</span><span class="n">nd</span><span class="o">.</span><span class="n">array</span><span class="p">(</span><span class="n">data</span><span class="p">,</span><span class="n">ctx</span> <span class="o">=</span> <span class="n">mx</span><span class="o">.</span><span class="n">gpu</span><span class="p">(</span><span class="mi">0</span><span class="p">)))</span>
</pre></div>
</div>
<p>Defined in src/operator/contrib/fft.cc:L56</p>
<table class="docutils field-list" frame="void" rules="none">
<col class="field-name"/>
<col class="field-body"/>
<tbody valign="top">
<tr class="field-odd field"><th class="field-name">Parameters:</th><td class="field-body"><ul class="first simple">
<li><strong>data</strong> (<a class="reference internal" href="ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – Input data to the FFTOp.</li>
<li><strong>compute_size</strong> (<em>int, optional, default='128'</em>) – Maximum size of sub-batch to be forwarded at one time</li>
<li><strong>out</strong> (<em>NDArray, optional</em>) – The output NDArray to hold the result.</li>
</ul>
</td>
</tr>
<tr class="field-even field"><th class="field-name">Returns:</th><td class="field-body"><p class="first"><strong>out</strong> –
The output of this function.</p>
</td>
</tr>
<tr class="field-odd field"><th class="field-name">Return type:</th><td class="field-body"><p class="first last">NDArray or list of NDArrays</p>
</td>
</tr>
</tbody>
</table>
</dd></dl>
<dl class="function">
<dt id="mxnet.ndarray.contrib.ifft">
<code class="descclassname">mxnet.ndarray.contrib.</code><code class="descname">ifft</code><span class="sig-paren">(</span><em>data=None</em>, <em>compute_size=_Null</em>, <em>out=None</em>, <em>name=None</em>, <em>**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.ndarray.contrib.ifft" title="Permalink to this definition"></a></dt>
<dd><p>Apply 1D ifft to input”</p>
<div class="admonition note">
<p class="first admonition-title">Note</p>
<p class="last"><cite>ifft</cite> is only available on GPU.</p>
</div>
<p>Currently accept 2 input data shapes: (N, d) or (N1, N2, N3, d). Data is in format: [real0, imag0, real1, imag1, ...].
Last dimension must be an even number.
The output data has shape: (N, d/2) or (N1, N2, N3, d/2). It is only the real part of the result.</p>
<p>Example:</p>
<div class="highlight-python"><div class="highlight"><pre><span></span><span class="n">data</span> <span class="o">=</span> <span class="n">np</span><span class="o">.</span><span class="n">random</span><span class="o">.</span><span class="n">normal</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="mi">3</span><span class="p">,</span><span class="mi">4</span><span class="p">))</span>
<span class="n">out</span> <span class="o">=</span> <span class="n">mx</span><span class="o">.</span><span class="n">contrib</span><span class="o">.</span><span class="n">ndarray</span><span class="o">.</span><span class="n">ifft</span><span class="p">(</span><span class="n">data</span> <span class="o">=</span> <span class="n">mx</span><span class="o">.</span><span class="n">nd</span><span class="o">.</span><span class="n">array</span><span class="p">(</span><span class="n">data</span><span class="p">,</span><span class="n">ctx</span> <span class="o">=</span> <span class="n">mx</span><span class="o">.</span><span class="n">gpu</span><span class="p">(</span><span class="mi">0</span><span class="p">)))</span>
</pre></div>
</div>
<p>Defined in src/operator/contrib/ifft.cc:L58</p>
<table class="docutils field-list" frame="void" rules="none">
<col class="field-name"/>
<col class="field-body"/>
<tbody valign="top">
<tr class="field-odd field"><th class="field-name">Parameters:</th><td class="field-body"><ul class="first simple">
<li><strong>data</strong> (<a class="reference internal" href="ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – Input data to the IFFTOp.</li>
<li><strong>compute_size</strong> (<em>int, optional, default='128'</em>) – Maximum size of sub-batch to be forwarded at one time</li>
<li><strong>out</strong> (<em>NDArray, optional</em>) – The output NDArray to hold the result.</li>
</ul>
</td>
</tr>
<tr class="field-even field"><th class="field-name">Returns:</th><td class="field-body"><p class="first"><strong>out</strong> –
The output of this function.</p>
</td>
</tr>
<tr class="field-odd field"><th class="field-name">Return type:</th><td class="field-body"><p class="first last">NDArray or list of NDArrays</p>
</td>
</tr>
</tbody>
</table>
</dd></dl>
<dl class="function">
<dt id="mxnet.ndarray.contrib.quantize">
<code class="descclassname">mxnet.ndarray.contrib.</code><code class="descname">quantize</code><span class="sig-paren">(</span><em>input=None</em>, <em>min_range=None</em>, <em>max_range=None</em>, <em>out_type=_Null</em>, <em>out=None</em>, <em>name=None</em>, <em>**kwargs</em><span class="sig-paren">)</span><a class="headerlink" href="#mxnet.ndarray.contrib.quantize" title="Permalink to this definition"></a></dt>
<dd><p>Quantize a input tensor from float to <cite>out_type</cite>,
with user-specified <cite>min_range</cite> and <cite>max_range</cite>.</p>
<p>[min_range, max_range] are scalar floats that spcify the range for
the input data. Each value of the tensor will undergo the following:</p>
<p><cite>out[i] = (in[i] - min_range) * range(OUTPUT_TYPE) / (max_range - min_range)</cite></p>
<p>here <cite>range(T) = numeric_limits<T>::max() - numeric_limits<T>::min()</cite></p>
<p>Defined in src/operator/contrib/quantize.cc:L41</p>
<table class="docutils field-list" frame="void" rules="none">
<col class="field-name"/>
<col class="field-body"/>
<tbody valign="top">
<tr class="field-odd field"><th class="field-name">Parameters:</th><td class="field-body"><ul class="first simple">
<li><strong>input</strong> (<a class="reference internal" href="ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – A ndarray/symbol of type <cite>float32</cite></li>
<li><strong>min_range</strong> (<a class="reference internal" href="ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – The minimum scalar value possibly produced for the input</li>
<li><strong>max_range</strong> (<a class="reference internal" href="ndarray.html#mxnet.ndarray.NDArray" title="mxnet.ndarray.NDArray"><em>NDArray</em></a>) – The maximum scalar value possibly produced for the input</li>
<li><strong>out_type</strong> (<em>{'uint8'},optional, default='uint8'</em>) – Output data type.</li>
<li><strong>out</strong> (<em>NDArray, optional</em>) – The output NDArray to hold the result.</li>
</ul>
</td>
</tr>
<tr class="field-even field"><th class="field-name">Returns:</th><td class="field-body"><p class="first"><strong>out</strong> –
The output of this function.</p>
</td>
</tr>
<tr class="field-odd field"><th class="field-name">Return type:</th><td class="field-body"><p class="first last">NDArray or list of NDArrays</p>
</td>
</tr>
</tbody>
</table>
</dd></dl>
<script>auto_index("api-reference");</script></div>
</div>
</div>
</div>
<div aria-label="main navigation" class="sphinxsidebar rightsidebar" role="navigation">
<div class="sphinxsidebarwrapper">
<h3><a href="../../../index.html">Table Of Contents</a></h3>
<ul>
<li><a class="reference internal" href="#">Contrib NDArray API</a><ul>
<li><a class="reference internal" href="#overview">Overview</a></li>
<li><a class="reference internal" href="#contrib">Contrib</a></li>
<li><a class="reference internal" href="#api-reference">API Reference</a></li>
</ul>
</li>
</ul>
</div>
</div>
</div><div class="footer">
<div class="section-disclaimer">
<div class="container">
<div>
<img height="60" src="https://raw.githubusercontent.com/dmlc/web-data/master/mxnet/image/apache_incubator_logo.png"/>
<p>
Apache MXNet is an effort undergoing incubation at The Apache Software Foundation (ASF), <strong>sponsored by the <i>Apache Incubator</i></strong>. Incubation is required of all newly accepted projects until a further review indicates that the infrastructure, communications, and decision making process have stabilized in a manner consistent with other successful ASF projects. While incubation status is not necessarily a reflection of the completeness or stability of the code, it does indicate that the project has yet to be fully endorsed by the ASF.
</p>
<p>
"Copyright © 2017-2018, The Apache Software Foundation
Apache MXNet, MXNet, Apache, the Apache feather, and the Apache MXNet project logo are either registered trademarks or trademarks of the Apache Software Foundation."
</p>
</div>
</div>
</div>
</div> <!-- pagename != index -->
</div>
<script crossorigin="anonymous" integrity="sha384-0mSbJDEHialfmuBBQP6A4Qrprq5OVfW37PRR3j5ELqxss1yVqOtnepnHVP9aJ7xS" src="https://maxcdn.bootstrapcdn.com/bootstrap/3.3.6/js/bootstrap.min.js"></script>
<script src="../../../_static/js/sidebar.js" type="text/javascript"></script>
<script src="../../../_static/js/search.js" type="text/javascript"></script>
<script src="../../../_static/js/navbar.js" type="text/javascript"></script>
<script src="../../../_static/js/clipboard.min.js" type="text/javascript"></script>
<script src="../../../_static/js/copycode.js" type="text/javascript"></script>
<script src="../../../_static/js/page.js" type="text/javascript"></script>
<script type="text/javascript">
$('body').ready(function () {
$('body').css('visibility', 'visible');
});
</script>
</body>
</html>