Opencv 2.2 Released

[GoogleCodeExporter]

What steps will reproduce the problem?
New release of library backend.

What is the expected output? What do you see instead?

What version of the product are you using? On what operating system?

Please provide any additional information below.
Opencv 2.2 has been released with a big reorganization, do you know when 
pyopencv might transition to the new library?

Original issue reported on code.google.com by aber...@gmail.com on 6 Dec 2010 at 3:13

[GoogleCodeExporter]

Hi Attila,

Thanks for raising the issue. At the moment, I have not got any plan to develop 
PyOpenCV for OpenCV 2.2 yet. I know it is not so difficult to migrate to the 
new version. It will only take less than a week to do so. However, I am 
currently busy with a project which does not use OpenCV. In fact, I have 
replaced OpenCV with Eigen in this project. Also, I would like to hear more 
feedbacks from the community about WG's new Python wrapper for OpenCV 2.2. If 
it is not as good as the community expects, then I will devote some time to 
develop PyOpenCV for OpenCV 2.2.

Cheers,
Minh-Tri

Original comment by pmtri80@gmail.com on 9 Dec 2010 at 9:21

• Changed state: Accepted

[GoogleCodeExporter]

Hi!

I have tried to use OpenCV 2.2 with its newest Python wrapper; however after a 
bit i found those bindings really limited (machine learning isn't implemented, 
many functions are missing): the code is bugged (some examples doesn't work!) 
and the documentation isn't updated, so working with it in Python is almost 
impossible.

Now I tried PyOpenCV with OpenCV 2.1 and Python 2.7; very well implemented, 
fast, and easy to use. I'm sure I'll love it.

Thanks for your efforts, I hope you're going to migrate to the new version, 
cause OpenCV2.2 needs you :) (Sorry for my poor english)

Cheers,
Leonard

Original comment by RockerL...@gmail.com on 19 Jan 2011 at 12:31

[GoogleCodeExporter]

Hi Minh-Tri,

I also tried the new bindings.... Some things are nice about them, but I agree 
that they are (far) less complete then pyopencv and their documentation sucks.  
On the other hand the new python bindings seem automatically generated by 
parsing the C source code.  I wonder whether it might be possible to somehow 
combine these two efforts and build one set of bindings that are actually fully 
on par with the C++ interface.

Also:  I noticed on the updates that the OpenCV development team actually 
looked at pyopencv when considering what to do about their python bindings, and 
still decided to build their own.  Do you know why?

Cheers:
Attila

Original comment by aber...@gmail.com on 19 Jan 2011 at 3:02

[GoogleCodeExporter]

Hi Leonard and Attila,

Thanks very much for your feedbacks. They are very encouraging for me to 
continue developing PyOpenCV.

The PyOpenCV project started with using Py++ so that new bindings can be 
automatically generated by parsing the header files. However, soon I realized 
that doing that way alone would still bring a lot of inconveniences to the user 
because it is virtually unpythonic. There were quite a number of coding styles 
in OpenCV 2.1 and earlier versions. Eventually, I wrote a number of templates 
to deal with those coding styles. It is those templates that free the user from 
low-level issues that a C/C++ programmer usually has to deal with, like 
pointers, unbounded arrays, and garbage collection. 

Technically, I can just run the current generating code of PyOpenCV on OpenCV 
2.2's header files to create new bindings. However, I would rather spend more 
time to determine the functionalities of new functions and their arguments, and 
decide how to expose them in such a way most convenient to the user, and write 
a new template to expose if necessary. After all, I am also a user of PyOpenCV 
and I want to free myself from low-level C stuff, too.

As for the reason why they decided to build their own bindings after looking at 
pyopencv, I don't know. I guess one reason is that they just don't want OpenCV 
to be involved with Boost. I may be wrong though.

Cheers,
Minh-Tri

Original comment by pmtri80@gmail.com on 19 Jan 2011 at 8:35

[GoogleCodeExporter]

Hi Minh-Tri,

first of all, I wanted to thank you that you started PyOpenCV: its inclusion in 
Python(x,y) first made me aware that OpenCV is actually available under Python!

Since OpenCV seems to be stunningly powerful, I have tried to work with it a 
bit. However, the syntax of PyOpenCV seems to be much less compatible to the 
original C-version (e.g. by using imshow, instead of the ShowImage from the 
standard OpenCV). Being new to the field and not knowing where to start, I 
eventually installed Python 2.7 and WG's OpenCV 2.2 - only to find that under 
Win7 already the "NamedWindow" command is buggy :(

At this point I am not experienced enough to contribute with the porting. 
However, I would be very grateful if you could bring OpenCV2.2 PyOpenCV: then 
it would be possible to learn OpenCV from the O'Reilly book, and use the 
convenience of Python for it.

If you can do it, that would be great. Otherwise, thanks for the work that you 
have already done.

cheers
thomas

Original comment by thomas.h...@gmail.com on 22 Feb 2011 at 3:20

[GoogleCodeExporter]

OpenCV2.2 on Python3.x would seem to be an excellent combination.  

Numpy is now at 3.1 (any day releasing 3.2) and Boost is available up to 
Python3.2 from http://www.lfd.uci.edu/~gohlke/pythonlibs/.  That means it's 
only pyopencv that's lagging.

I'm just starting with Python and would like to avoid learning anything about 
Python2.x

I would also like to forget the hours that I have spent trying to get Cmake to 
find Python bits and pieces - their 'directions' are hopelessly inadequate.  On 
a related note, things like IPP v7 still have not been updated in the CMake 
files ???

Would be great if you could release a complete implementation of OpenCV2.2 on 
Python3.x.  I think at that point the OpenCV team would see the superiority of 
your approach and adopt pyopencv (if you let them).

Other people have said that your efforts are important to Python and OpenCV, 
and I'm in full agreement.  I think programmers like you are what the open 
source world depends on.

Original comment by dmy...@gmail.com on 17 Mar 2011 at 6:53