From: andrew mcelroy <sophrinix@...>
Date: 2012-09-28T23:25:35+09:00
Subject: Re: image processing

--20cf307f31e22d496604cac3d4a7
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable

On Mon, Sep 24, 2012 at 8:40 PM, Nicol=F2 L. <lists@ruby-forum.com> wrote:

> I'm trying to build a program meant to recognize plants by looking at a
> picture of their leafs or similar parts.
>

I'm doing this kind of detection with animals.
There is a ruby opencv library, but it doesn't have 100% api coverage and
opencv can be a beast in its own right to install.


>
> The approach that I had thought to would be:
> - create a mathematical model of the leaves to be recognized
> - maybe use some algorithm to combine two pictures from different angles
> into one with depth measurement (absolutely don't know how this is done)
> - confront the elements of the picture using distance or colors with the
> models in the database (basically using only the shape of the borders)
>

Stone Age. You will need a lot of samples, but there are a lot of
algorithms you aren't even considering.
You need to look more into Boosting, Bagging, and Stacking algorithms.
For starters look at AdaBoost or WaldBoost algorithm.
Also, check out Bag of Words and SIFT.


>
> Therefore I am looking for some good and deep guides/sources to teach me
> something about this kind of "data modeling", geometrical analysis and
> image processing. For now the best that I could find is the Camelia
> Library (witch helps a little with colors, but I don't think will be
> enough for the shape detection).
>

You need something like opencv.  I think the library you need is
ruby-opencv.
It can be found on github. Follow the instructions for installing opencv
with python first
and build it from source. Also, you are going to want to have fftw library
installed.

As for reading material, computer vision/ machine learning for plants has a
lot of literature.
Start using scholar.google.com and don't be afraid to ask people at other
.edu addresses for advise.

hope this helps.

Andrew McElroy

>
>
> I'm aware that this is not a simple project, but I am willing to learn
> and spend a good amount of months to get this started
>
> --
> Posted via http://www.ruby-forum.com/.
>
>

--20cf307f31e22d496604cac3d4a7
Content-Type: text/html; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable

<br><br><div class=3D"gmail_quote">On Mon, Sep 24, 2012 at 8:40 PM, Nicol=
=F2 L. <span dir=3D"ltr">&lt;<a href=3D"mailto:lists@ruby-forum.com" target=
=3D"_blank">lists@ruby-forum.com</a>&gt;</span> wrote:<br><blockquote class=
=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1px #ccc solid;padd=
ing-left:1ex">
I&#39;m trying to build a program meant to recognize plants by looking at a=
<br>
picture of their leafs or similar parts.<br></blockquote><div><br></div><di=
v>I&#39;m doing this kind of detection with animals.</div><div>There is a r=
uby opencv library, but it doesn&#39;t have 100% api coverage and opencv ca=
n be a beast in its own right to install.</div>
<div>=A0</div><blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;=
border-left:1px #ccc solid;padding-left:1ex">
<br>
The approach that I had thought to would be:<br>
- create a mathematical model of the leaves to be recognized<br>
- maybe use some algorithm to combine two pictures from different angles<br=
>
into one with depth measurement (absolutely don&#39;t know how this is done=
)<br>
- confront the elements of the picture using distance or colors with the<br=
>
models in the database (basically using only the shape of the borders)<br><=
/blockquote><div><br></div><div>Stone Age. You will need a lot of samples, =
but there are a lot of algorithms you aren&#39;t even considering.</div>
<div>You need to look more into Boosting, Bagging, and Stacking=A0algorithm=
s.</div><div>For starters look at AdaBoost or WaldBoost=A0algorithm.</div><=
div>Also, check out Bag of Words and SIFT.</div><div>=A0</div><blockquote c=
lass=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1px #ccc solid;=
padding-left:1ex">

<br>
Therefore I am looking for some good and deep guides/sources to teach me<br=
>
something about this kind of &quot;data modeling&quot;, geometrical analysi=
s and<br>
image processing. For now the best that I could find is the Camelia<br>
Library (witch helps a little with colors, but I don&#39;t think will be<br=
>
enough for the shape detection).<br></blockquote><div><br></div><div>You ne=
ed something like opencv. =A0I think the library you need is ruby-opencv.</=
div><div>It can be found on github. Follow the instructions for installing =
opencv with python first</div>
<div>and build it from source. Also, you are going to want to have fftw lib=
rary installed.</div><div><br></div><div>As for reading material, computer =
vision/ machine learning for plants has a lot of literature.</div><div>
Start using <a href=3D"http://scholar.google.com">scholar.google.com</a> an=
d don&#39;t be afraid to ask people at other .edu addresses for advise.</di=
v><div><br></div><div>hope this helps.</div><div><br></div><div>Andrew McEl=
roy</div>
<blockquote class=3D"gmail_quote" style=3D"margin:0 0 0 .8ex;border-left:1p=
x #ccc solid;padding-left:1ex">
<br>
<br>
I&#39;m aware that this is not a simple project, but I am willing to learn<=
br>
and spend a good amount of months to get this started<br>
<span class=3D"HOEnZb"><font color=3D"#888888"><br>
--<br>
Posted via <a href=3D"http://www.ruby-forum.com/" target=3D"_blank">http://=
www.ruby-forum.com/</a>.<br>
<br>
</font></span></blockquote></div><br>

--20cf307f31e22d496604cac3d4a7--