D7net Mini Sh3LL v1
| Current File : /var/www/html/../html/cvprlab/code/index.xml |
<?xml version="1.0" encoding="utf-8" standalone="yes" ?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
<channel>
<title>Codes | CVPR Lab</title>
<link>/code/</link>
<atom:link href="/code/index.xml" rel="self" type="application/rss+xml" />
<description>Codes</description>
<generator>Source Themes Academic (https://sourcethemes.com/academic/)</generator><language>en-us</language>
<image>
<url>/img/logo.png</url>
<title>Codes</title>
<link>/code/</link>
</image>
<item>
<title>Moving Object Detection Sequences</title>
<link>/code/modse/</link>
<pubDate>Mon, 01 Jan 0001 00:00:00 +0000</pubDate>
<guid>/code/modse/</guid>
<description><p><strong>Sequence MSA</strong></p>
<p>Sequence MSA is a home-made indoor sequence manually labeled, consisting of 528 frames of 320*240 spatial resolution, acquired at a frequency of 30 fps. The scene consists of a university hall, where a man comes in, leaves a bag on the floor, and then comes out. It represents an example of easy sequence, in that lighting conditions are quite stable and moving objects are well contrasted with the background (there is no camouflage); however, strong shadows cast by moving objects can be observed in the entire sequence.</p>
<p>Some snapshots:</p>
<table>
<thead>
<tr>
<th></th>
<th></th>
</tr>
</thead>
<tbody>
<tr>
<td><img src="/img/download.gif" alt="download"></td>
<td>Download the sequence <a href="/rar/MSA.rar">here</a> (27.2MB rar file containing 528 PPM images).</td>
</tr>
<tr>
<td><img src="/img/download.gif" alt="download"></td>
<td>Download the Ground Truth <a href="/txt/GroundTruth.txt">here</a> (txt file containing, for each sequence frame, number of moving objects and their bounding box coordinates).</td>
</tr>
</tbody>
</table>
<p>\</p>
<table>
<thead>
<tr>
<th>Please cite:</th>
</tr>
</thead>
<tbody>
<tr>
<td>L. Maddalena, A. Petrosino, <a href="https://link.springer.com/chapter/10.1007%2F978-3-540-87536-9_67">A Self-Organizing Approach to Background Subtraction for Visual Surveillance Applications</a>, <strong>IEEE Transactions on Image Processing</strong>, vol. 17, no. 7, 1168-1177, 2008.</td>
</tr>
</tbody>
</table>
<!-- raw HTML omitted -->
</description>
</item>
<item>
<title>Moving Object Detection Software</title>
<link>/code/mods/</link>
<pubDate>Mon, 01 Jan 0001 00:00:00 +0000</pubDate>
<guid>/code/mods/</guid>
<description><p>Implementation of SOBS algorithm as described in:</p>
<p>L. Maddalena, A. Petrosino, <a href="https://ieeexplore.ieee.org/document/4527178/">A Self-Organizing Approach to Background Subtraction for Visual Surveillance Applications</a>, <strong>IEEE Transactions on Image Processing</strong>, DOI: 10.1109/TIP.2008.924285, Vol. 17, no. 7, pagg. 1168-1177, July 2008</p>
<p><strong>Usage</strong>:</p>
<!-- raw HTML omitted -->
<p>where</p>
<ul>
<li><code>&lt;SeqName&gt;</code>: sequence name (complete path), not including frame numbers. Image sequences consist of binary PPM image frames with consecutive numbers, named in the following form</li>
</ul>
<!-- raw HTML omitted -->
<p>The number of digits for <!-- raw HTML omitted --> must be the same for all sequence frames (e.g. a sequence with 120 frames must be numbered from 1001 to 1120, and not from 1 to 120)</p>
<ul>
<li>
<p><code>&lt;#FirstFrame&gt;, &lt;#LastFrame&gt;</code>: number of first and last sequence frame to be considered.</p>
</li>
<li>
<p>[parameters]: optional, including:</p>
<ul>
<li>
<p>-n #: (square root of) number of weight vectors for each pixel. Default 3</p>
</li>
<li>
<p>-K #: Number of initial frames for calibration. Default 200</p>
</li>
<li>
<p>-e1 #: Distance threshold e1 for calibration phase (eqn. (2)). Default 0.1</p>
</li>
<li>
<p>-e2 #: Distance threshold e2 for online phase (eqn. (2)). Default 0.03</p>
</li>
<li>
<p>-c1 #: Learning rate c1 for calibration phase (eqn. (4)). Default 1.0</p>
</li>
<li>
<p>-c2 #: Learning rate c2 for online phase (eqn. (4)). Default 0.05</p>
</li>
<li>
<p>-g #: Value for g in eqn. (5). Default 0.7</p>
</li>
<li>
<p>-b #: Value for b in eqn. (5). Default 1.0</p>
</li>
<li>
<p>-tS #: Value for tS in eqn. (5). Default 0.1</p>
</li>
<li>
<p>-tH #: Value for tH in eqn. (5). Default 10.0</p>
</li>
<li>
<p>-s: To apply shadow removal. Default: no shadow removal</p>
</li>
<li>
<p>-m: To save background model images. Default: do not save</p>
</li>
<li>
<p>-l: To save just last detection mask. Default: save all</p>
</li>
<li>
<p><code>&lt;!--[if !supportEmptyParas]--&gt; &lt;!--[endif]--&gt;</code></p>
</li>
</ul>
</li>
</ul>
<p><strong>Example of use:</strong></p>
<!-- raw HTML omitted -->
<p>where sequence WavingTrees, coming from sequences adopted in K. Toyama, J. Krumm, B. Brumitt, and B. Meyers, “Wallflower: principles and practice of background maintenance,” in Proc. 7th IEEE Conf. Computer Vision, 1999, vol. 1, pp. 255–261, has been saved in binary PPM image files named:</p>
<!-- raw HTML omitted -->
<p>and stored in directory c:/Sequences/WavingTrees.
<br>
\</p>
<table>
<thead>
<tr>
<th></th>
<th></th>
</tr>
</thead>
<tbody>
<tr>
<td><img src="/img/download.gif" alt="download"></td>
<td>Download the SOBS software <a href="/rar/SOBS.rar">here</a>.</td>
</tr>
</tbody>
</table>
<!-- raw HTML omitted -->
</description>
</item>
<item>
<title>Moving Object Detection Software: SC-SOBS</title>
<link>/code/sc-sobs/</link>
<pubDate>Mon, 01 Jan 0001 00:00:00 +0000</pubDate>
<guid>/code/sc-sobs/</guid>
<description><p>This page has been created in order to distribute a prototype software implementing the Spatially Coherent Self-Organizing Background Subtraction (SC-SOBS) algorithm presented in</p>
<p><strong>L. Maddalena, A. Petrosino, The SOBS Algorithm: What Are the Limits?, IEEE Workshop on Change Detection, at CVPR 2012</strong></p>
<p>If you use the software, please cite the above mentioned paper.</p>
<p>Click <a href="/zip/SC-SOBSforCVPR2012.zip">here</a> to download the Windows executable (WinZip compressed) together with the needed OpenCV .dll's.
Basic usage for Change Detection Competition: Unzip the SC-SOBS.exe and the .dll's into the directory holding the &ldquo;dataset&rdquo; directory containing the Change Detection Challenge sequences [1]. To obtain all masks for the &ldquo;baseline/highway&rdquo; sequence, just type:</p>
<p>SC-SOBS dataset/baseline/highway/input/in</p>
<p>Click here to download the masks computed by SC-SOBS for the whole dataset.
Usage with generic image sequences:</p>
<p>SC-SOBS <code>&lt;SeqName&gt; &lt;#FirstFrame&gt; &lt;#LastFrame&gt; [Parameters]</code></p>
<p>where</p>
<ul>
<li>
<p><code>&lt;SeqName&gt;</code>: sequence name (complete path), not including frame numbers. Image sequences consist of binary .jpg image frames with consecutive 6 digit numbers, named in the following form <code>&lt;SeqName&gt;&lt;number&gt;.jpg</code></p>
</li>
<li>
<p><code>&lt;#FirstFrame&gt;</code>: number of first sequence frame to be considered. Default: 1</p>
</li>
<li>
<p><code>&lt;#LastFrame&gt;</code>: number of last sequence frame to be considered. Default: toIdx (toIdx read from file ‘temporalROI.txt’ as in [1])</p>
</li>
<li>
<p><code>[parameters]</code>: optional, including:</p>
<ul>
<li>
<p>-n #: (square root of) number of weight vectors for each pixel. Default: 3</p>
</li>
<li>
<p>-K #: Number of initial frames for training. Default: fromIdx-1 (fromIdx read from file ‘temporalROI.txt’ as in [1])</p>
</li>
<li>
<p>-e1 #: Distance threshold e1 for training phase (Eq. (12)). Default: 1.0</p>
</li>
<li>
<p>-e2 #: Distance threshold e2 for testing phase (Eq. (2)). Default: 0.008</p>
</li>
<li>
<p>-c1 #: Learning rate c1 for training phase (Eq. (14)). Default: 1.0</p>
</li>
<li>
<p>-c2 #: Learning rate c2 for testing phase (Eq. (14)). Default: 0.05</p>
</li>
<li>
<p>-Cw #: Size of the neighbourhood for Spatial Coherence (Eq. (10)). Default: 5</p>
</li>
<li>
<p>-s #: To apply shadow removal (as in [2]). Default: 1 (apply)</p>
</li>
<li>
<p>-g #: Shadow detection value for g in eqn. (5) in [2]. Default: 0.7</p>
</li>
<li>
<p>-b #: Shadow detection value for b in eqn. (5) in [2]. Default: 1.0</p>
</li>
<li>
<p>-tS #: Shadow detection value for tS in eqn. (5) in [2]. Default: 0.1</p>
</li>
<li>
<p>-tH #: Shadow detection value for tH in eqn. (5) in [2]. Default: 10.0</p>
</li>
<li>
<p>-ROI #: To use ROI.bmp mask as in [1]. Default: 1 (do use)</p>
</li>
<li>
<p>-Med #: Size of the neighbourhood for Median Filtering Post-Processing. Default: 0 (no Post-Processing)</p>
</li>
<li>
<p>-m #: To save background model images. Default: 0 (do not save; only models for frames K-1 and #LastFrame are saved)</p>
</li>
<li>
<p>-l #: To save only last detection mask. Default 0 (save all in the temporal ROI)</p>
</li>
</ul>
</li>
</ul>
<p>Examples of use with generic image sequences:</p>
<ol>
<li>SC-SOBS</li>
</ol>
<p>Provides the above information on usage.</p>
<ol start="2">
<li>SC-SOBS c:/Sequences/WavingTrees/WavingTrees 1000 1247 -K 200 -e1 0.1 -e2 0.03 -c1 1.0 -c2 0.05 –l 1 –ROI 0</li>
</ol>
<p>where sequence WavingTrees, coming from sequences adopted in K. Toyama, J. Krumm, B. Brumitt, and B. Meyers, “Wallflower: principles and practice of background maintenance,” in Proc. 7th IEEE Conf. Computer Vision, 1999, vol. 1, pp. 255–261, has been saved in binary .jpg image files named:<br>
WavingTrees001000.jpg, …, WavingTrees001247.jpg</p>
<p>and stored in directory c:/Sequences/WavingTrees.
This gives the moving object detection mask for last frame (named bin001247.png) as well the background model (named Model001199.ppm) achieved by training on the first 200 frames and the updated background model (named Model001247.ppm) for the last frame.<br>
3) SC-SOBS c:/Sequences/WavingTrees/WavingTrees 1000 1247 -K 200 -e1 0.1 -e2 0.03 -c1 1.0 -c2 0.05 –l 1 –ROI 0 –Med 3</p>
<p>same as before, but applying median filtering post-processing in a 3x3 neighbourhood to masks (through OpenCV function cvSmooth).</p>
<!-- raw HTML omitted -->
</description>
</item>
<item>
<title>ROSE (Rough Outlier Set Extraction)</title>
<link>/code/rose/</link>
<pubDate>Mon, 01 Jan 0001 00:00:00 +0000</pubDate>
<guid>/code/rose/</guid>
<description><p>ROSE is a machine learning algorithm that implements outlier detection of an unlabeled spatiotemporal dataset using a rough set approach. It also provides a representative subset of the original data, describing the same structure, with which it is possible to detect the same outliers, named kernel set.</p>
<p>The paper that reports all the details and should be cited when the code is used is</p>
<p>Albanese A, Sankar K P, Petrosino A., <a href="https://ieeexplore.ieee.org/document/6365186/?arnumber=6365186">IEEE Transactions on Knowledge and Data Engineering</a>, Vol. 26, no. 1, pp. 194-207, 2014, DOI: 10.1109/TKDE.2012.234.</p>
<p>ROSE is written in Java and can be downloaded <a href="/tar/ROSEv3.0.tar.gz">here</a>. It is platform independent and bundled as jar package. Java Runtime Environment (JRE) is needed to run this software.</p>
<h3 id="rose-package-content">ROSE package content:</h3>
<ul>
<li>ROSE.jar (java executable file)</li>
<li>README.md</li>
<li>LICENSE</li>
<li>sample_dataset.txt</li>
<li>Original paper 2014_Albanese_Sankar_Petrosino.pdf</li>
<li>Supplementary material folder ttk2014010194</li>
</ul>
<p>ROSE can be executed from a Unix-like shell, a Terminal for Mac-OS or Windows CMD as follows:</p>
<p><code>java -jar ROSE.jar &lt;path input dataset&gt; &lt;number of outliers&gt; &lt;number of neighbors&gt; &lt;number of elements once&gt; &lt;alpha&gt;</code></p>
<p>This is an example using the bundled sample dataset (assuming you are in the ROSE root folder):</p>
<p><code>java -jar ROSE.jar sample_dataset.txt 16 16 18 1</code></p>
<h3 id="input-parameters">Input parameters</h3>
<ul>
<li>path input dataset - Input dataset filename (complete absolute path). This has to be a text file where each row contains numerical values, ranging in [0, 1] (normalized). Each numeric value must be space separated (two or three space characters are allowed). These values are the coordinates of the instances to be processed.<br>
Each row specifies:
<ul>
<li>
<p><strong>required configuration:</strong> <em>(x,y,t)</em> triplets where <em>(x,y)</em> is a spatial coordinate and (t) is a timestamp.<br>
Example:<br>
0.84506904 0.10248589 0.15307587<br>
0.84543708 0.10339779 0.15308264<br>
0.84524589 0.10418321 0.15308941<br>
0.84540431 0.10549813 0.15309618<br>
0.84588570 0.10667086 0.15310294\</p>
</li>
<li>
<p><strong>optional configuration</strong>: 13 values are allowed at most, (x,y,t) + (f1,&hellip;,f10) where the first 3 values are inherited from required configuration and <em>(f1,&hellip;,f10)</em> is a list of other features.<br>
Example:<br>
0.84506904 0.10248589 0.15307587 0.84588570 0.10667086 0.15310294<br>
0.84543708 0.10339779 0.15308264 0.84524589 0.10418321 0.15308941<br>
0.84540431 0.10549813 0.15309618 0.84588570 0.10667086 0.15310294\</p>
</li>
<li>
<p><strong>standard configuration</strong>: not all 13 values need to be specified and just some features are specified (the sample dataset only contains <em>(x,y,t)</em>).</p>
</li>
</ul>
</li>
<li><strong>number of outliers</strong> - the number of outliers to be extracted from the data set.</li>
<li><strong>number of neighbors</strong> - the number of neighbors to be considered.</li>
<li><strong>number of elements once</strong> - it is the number of items that are extracted from the input data set at each iteration (and then compared with the rest of the file). This must be greater than the number of outliers. number of elements once &gt; number of outliers</li>
<li><strong>alpha</strong> - is a multiplier value of the linear combination of the weights, ranging between [0,1] (for spatial weight) and 1 - alpha (for temporal weight).
Special cases: alpha = 1 only spatial components are considered. alpha = 0 only temporal component is considered.</li>
</ul>
<h3 id="output-data">Output data</h3>
<p>Each execution of the ROSE software will produce an output folder named as the dataset input file name. Each execution on the same input dataset will overwrite the existing output folder.</p>
<ul>
<li><strong>TopOutlier</strong>-<em><strong>numoutlier-alpha-weight-milliseconds.txt</strong></em> - (<em>milliseconds</em> value in the filename is used to maintain processing history, TopOutlier-<em>inputfilename-numoutliers-alpha-weight.txt</em> is the most recent iteration). These files contain TopOutliers search results at each iteration. The last two files contain the Lower and Upper approximation of the outlier set respectively.</li>
<li><strong>WeightedData</strong>-<em><strong>inputfilename-numoutliers-weight.txt</strong></em> - contains the starting points set with weights for each point.</li>
<li><strong>SolvingSet</strong>-<em><strong>milliseconds.txt</strong></em> - contains the kernel set (milliseconds value in the filename is used to maintain processing history, SolvingSet.txt is the most recent iteration).</li>
<li><strong>NegativeRegion.txt</strong> - contains discarded instances.
<br>
\</li>
</ul>
<!-- raw HTML omitted -->
<hr>
</description>
</item>
</channel>
</rss>
AnonSec - 2021 | Recode By D7net