<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"
	xmlns:content="http://purl.org/rss/1.0/modules/content/"
	xmlns:dc="http://purl.org/dc/elements/1.1/"
	xmlns:atom="http://www.w3.org/2005/Atom"
	xmlns:sy="http://purl.org/rss/1.0/modules/syndication/"
		>
<channel>
	<title>Comments on: &#8220;Fixing&#8221; the OpenCV&#8217;s implementation of Viola-Jones algorithm</title>
	<atom:link href="http://www.computer-vision-software.com/blog/2009/04/fixing-opencv/feed/" rel="self" type="application/rss+xml" />
	<link>http://www.computer-vision-software.com/blog/2009/04/fixing-opencv/</link>
	<description>Rhonda Ltd., computer vision software blog</description>
	<lastBuildDate>Mon, 06 Sep 2010 10:44:54 +0000</lastBuildDate>
	<sy:updatePeriod>hourly</sy:updatePeriod>
	<sy:updateFrequency>1</sy:updateFrequency>
	<generator>http://wordpress.org/?v=3.0.1</generator>
	<item>
		<title>By: Aleksey Kodubets</title>
		<link>http://www.computer-vision-software.com/blog/2009/04/fixing-opencv/comment-page-1/#comment-198</link>
		<dc:creator>Aleksey Kodubets</dc:creator>
		<pubDate>Mon, 10 Aug 2009 10:43:14 +0000</pubDate>
		<guid isPermaLink="false">http://www.computer-vision-software.com/blog/?p=41#comment-198</guid>
		<description>Roys,

Thank you for your feedback. It is very important for us to know that developement community is using our results.

Aleksey</description>
		<content:encoded><![CDATA[<p>Roys,</p>
<p>Thank you for your feedback. It is very important for us to know that developement community is using our results.</p>
<p>Aleksey</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: roys</title>
		<link>http://www.computer-vision-software.com/blog/2009/04/fixing-opencv/comment-page-1/#comment-197</link>
		<dc:creator>roys</dc:creator>
		<pubDate>Sun, 09 Aug 2009 14:36:37 +0000</pubDate>
		<guid isPermaLink="false">http://www.computer-vision-software.com/blog/?p=41#comment-197</guid>
		<description>Hi,
I tried to do the steps you cover here in your article, and added a few of my own.
I posted an article of my own about it:
http://www.morethantechnical.com/2009/08/09/near-realtime-face-detection-on-the-iphone-w-opencv-port-wcodevideo/

It also includes code, and a sample video.
Roy.</description>
		<content:encoded><![CDATA[<p>Hi,<br />
I tried to do the steps you cover here in your article, and added a few of my own.<br />
I posted an article of my own about it:<br />
<a href="http://www.morethantechnical.com/2009/08/09/near-realtime-face-detection-on-the-iphone-w-opencv-port-wcodevideo/" rel="nofollow">http://www.morethantechnical.com/2009/08/09/near-realtime-face-detection-on-the-iphone-w-opencv-port-wcodevideo/</a></p>
<p>It also includes code, and a sample video.<br />
Roy.</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: Aleksey Kodubets</title>
		<link>http://www.computer-vision-software.com/blog/2009/04/fixing-opencv/comment-page-1/#comment-150</link>
		<dc:creator>Aleksey Kodubets</dc:creator>
		<pubDate>Thu, 09 Jul 2009 10:23:26 +0000</pubDate>
		<guid isPermaLink="false">http://www.computer-vision-software.com/blog/?p=41#comment-150</guid>
		<description>Hi,

Personally I didn&#039;t agree with your comment . I understand that you would like to expect more significant reduce of execution time and at least saving quality on the same level. If so, then most probably you would read our other post:

http://www.computer-vision-software.com/blog/2009/06/fastfurious-face-detection-with-opencv/

where execution time is increased up to 2-3 times even without fixed point re-implementation of Viola-Jones algorithm, that implementation saves quality on the same level.

Let me know if you have any more question.

Aleksey</description>
		<content:encoded><![CDATA[<p>Hi,</p>
<p>Personally I didn&#8217;t agree with your comment . I understand that you would like to expect more significant reduce of execution time and at least saving quality on the same level. If so, then most probably you would read our other post:</p>
<p><a href="http://www.computer-vision-software.com/blog/2009/06/fastfurious-face-detection-with-opencv/" rel="nofollow">http://www.computer-vision-software.com/blog/2009/06/fastfurious-face-detection-with-opencv/</a></p>
<p>where execution time is increased up to 2-3 times even without fixed point re-implementation of Viola-Jones algorithm, that implementation saves quality on the same level.</p>
<p>Let me know if you have any more question.</p>
<p>Aleksey</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: cheap computers</title>
		<link>http://www.computer-vision-software.com/blog/2009/04/fixing-opencv/comment-page-1/#comment-149</link>
		<dc:creator>cheap computers</dc:creator>
		<pubDate>Thu, 09 Jul 2009 09:35:33 +0000</pubDate>
		<guid isPermaLink="false">http://www.computer-vision-software.com/blog/?p=41#comment-149</guid>
		<description>Its a useful software but their input in the overall execution time is very insignificant.</description>
		<content:encoded><![CDATA[<p>Its a useful software but their input in the overall execution time is very insignificant.</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: Aleksey Kodubets</title>
		<link>http://www.computer-vision-software.com/blog/2009/04/fixing-opencv/comment-page-1/#comment-74</link>
		<dc:creator>Aleksey Kodubets</dc:creator>
		<pubDate>Fri, 22 May 2009 00:28:38 +0000</pubDate>
		<guid isPermaLink="false">http://www.computer-vision-software.com/blog/?p=41#comment-74</guid>
		<description>We used 64-bit DWORD for saving square sums. As I remember calculation of integral image takes fixed time and it is 5-7% of all time. Hope it will useful.</description>
		<content:encoded><![CDATA[<p>We used 64-bit DWORD for saving square sums. As I remember calculation of integral image takes fixed time and it is 5-7% of all time. Hope it will useful.</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: Ching-Min Wang</title>
		<link>http://www.computer-vision-software.com/blog/2009/04/fixing-opencv/comment-page-1/#comment-73</link>
		<dc:creator>Ching-Min Wang</dc:creator>
		<pubDate>Thu, 21 May 2009 05:40:46 +0000</pubDate>
		<guid isPermaLink="false">http://www.computer-vision-software.com/blog/?p=41#comment-73</guid>
		<description>Hi, Aleksey,

Thanks for your reply. My target processor only supports 32-bit integer arithmetic, and amount of memory is also critical. A straightforword method to compute square integral image is to use 64-bit arithmetic. Compared with sum integral image, it would cost two times of memory. So I wonder if there is any smart and efficient method to calcaulte and store square integral image. 

CM</description>
		<content:encoded><![CDATA[<p>Hi, Aleksey,</p>
<p>Thanks for your reply. My target processor only supports 32-bit integer arithmetic, and amount of memory is also critical. A straightforword method to compute square integral image is to use 64-bit arithmetic. Compared with sum integral image, it would cost two times of memory. So I wonder if there is any smart and efficient method to calcaulte and store square integral image. </p>
<p>CM</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: Aleksey Kodubets</title>
		<link>http://www.computer-vision-software.com/blog/2009/04/fixing-opencv/comment-page-1/#comment-72</link>
		<dc:creator>Aleksey Kodubets</dc:creator>
		<pubDate>Wed, 20 May 2009 10:53:21 +0000</pubDate>
		<guid isPermaLink="false">http://www.computer-vision-software.com/blog/?p=41#comment-72</guid>
		<description>Hi Ching-Min Wang,

&quot;Also we replaced using of sqrt with its *integral* version.&#039; 

It is wrong comment from Sergey Kislov. He meant that He used fixed point sqrt function which is calculated in integer numbers. The calculation of square integral image was not changed essentially.

Aleksey</description>
		<content:encoded><![CDATA[<p>Hi Ching-Min Wang,</p>
<p>&#8220;Also we replaced using of sqrt with its *integral* version.&#8217; </p>
<p>It is wrong comment from Sergey Kislov. He meant that He used fixed point sqrt function which is calculated in integer numbers. The calculation of square integral image was not changed essentially.</p>
<p>Aleksey</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: Ching-Min Wang</title>
		<link>http://www.computer-vision-software.com/blog/2009/04/fixing-opencv/comment-page-1/#comment-71</link>
		<dc:creator>Ching-Min Wang</dc:creator>
		<pubDate>Tue, 19 May 2009 05:57:27 +0000</pubDate>
		<guid isPermaLink="false">http://www.computer-vision-software.com/blog/?p=41#comment-71</guid>
		<description>It is really a great job. Recently, I try to implement haar by fixed point. I had done some the observations as you done in the article. After analyzing some methods, I found it is hard to find a suitable one to compute integral of square of the image pixels. I found in your article in which you had mentioned:

&quot;Also we replaced using of sqrt with its integral version. This let us calculate variance in integer numbers&quot;. 

It might be the solution to the problem. But I do not know what it means. Could you give me some hints?


BR,

CM</description>
		<content:encoded><![CDATA[<p>It is really a great job. Recently, I try to implement haar by fixed point. I had done some the observations as you done in the article. After analyzing some methods, I found it is hard to find a suitable one to compute integral of square of the image pixels. I found in your article in which you had mentioned:</p>
<p>&#8220;Also we replaced using of sqrt with its integral version. This let us calculate variance in integer numbers&#8221;. </p>
<p>It might be the solution to the problem. But I do not know what it means. Could you give me some hints?</p>
<p>BR,</p>
<p>CM</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: Aleksey Kodubets</title>
		<link>http://www.computer-vision-software.com/blog/2009/04/fixing-opencv/comment-page-1/#comment-57</link>
		<dc:creator>Aleksey Kodubets</dc:creator>
		<pubDate>Mon, 27 Apr 2009 23:24:10 +0000</pubDate>
		<guid isPermaLink="false">http://www.computer-vision-software.com/blog/?p=41#comment-57</guid>
		<description>Hi Eyal,

The style of the reports are rather science with some confidential restrictions like Paul Viola, Michael Jones&#039;style they are working for Mitsubishi Electric Research Laboratories and don&#039;t share their code but share their results. So, our strategy are publish only results. Although in other post comments, the mans was able to repeat the same result with additional comments. 

Thanks,
Aleksey</description>
		<content:encoded><![CDATA[<p>Hi Eyal,</p>
<p>The style of the reports are rather science with some confidential restrictions like Paul Viola, Michael Jones&#8217;style they are working for Mitsubishi Electric Research Laboratories and don&#8217;t share their code but share their results. So, our strategy are publish only results. Although in other post comments, the mans was able to repeat the same result with additional comments. </p>
<p>Thanks,<br />
Aleksey</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: Eyal</title>
		<link>http://www.computer-vision-software.com/blog/2009/04/fixing-opencv/comment-page-1/#comment-56</link>
		<dc:creator>Eyal</dc:creator>
		<pubDate>Mon, 27 Apr 2009 11:52:24 +0000</pubDate>
		<guid isPermaLink="false">http://www.computer-vision-software.com/blog/?p=41#comment-56</guid>
		<description>Hi Sergey,

Great work!!! Is it possible to have the sources of your fixed point library.

Thanks,
Eyal</description>
		<content:encoded><![CDATA[<p>Hi Sergey,</p>
<p>Great work!!! Is it possible to have the sources of your fixed point library.</p>
<p>Thanks,<br />
Eyal</p>
]]></content:encoded>
	</item>
</channel>
</rss>
