I believe the Python dependency is just in wrapper generation--it wouldn't be in play at runtime.
On Wednesday, September 10, 2014 7:06:22 PM UTC-5, Max Suster wrote: > > > +1 > > I agree that it would be better to have at least a wrapper available with > python dependency than none at all. However, I am wondering though what is > the cost on speed of using this python dependency. A major feature that > makes me interested in Julia is the speed over languages like Ruby, > especially necessary for real-time image processing applications (high > frame rates). > > > On Wednesday, September 10, 2014 6:09:34 PM UTC+2, Kevin Squire wrote: >> >> While I also hope that native Julia solutions move forward, I would also >> find the OpenCV bindings quite useful, so +1! I also second the suggestion >> to use the existing python infrastructure. >> >> Cheers, >> Kevin >> >> On Wednesday, September 10, 2014, Boxiang Sun <[email protected]> wrote: >> >>> Hi Jake, >>> >>> Thanks for your comment! >>> >>> Use the exist Python header parser could save a lot of work. I will be >>> glad to accept it. If other friends believe the dependency of Python is not >>> a problem, too. Then I will write the the type conversion first(Mat, etc), >>> like Java binding did(../modules/java/generator). >>> >>> Regards, >>> Sun >>> >>> 在 2014年9月10日星期三UTC+8上午10时29分48秒，Jake Bolewski写道： >>>> >>>> Hi Boxiang, >>>> >>>> I'm excited that you are working on this. The one comment I would have >>>> is don't rewrite the custom header parser. It is maintained as part of >>>> the >>>> official build system in OpenCV and will be maintained. All the >>>> "official" >>>> OpenCV bindings depend on it. It is awkward to write python code to emit >>>> Julia code, but I feel you will save so much time in the end and have a >>>> chance of getting it into OpenCV core in the future if you reuse their >>>> infrastructure. >>>> >>>> Best, >>>> Jake >>>> >>>> >>>> On Tuesday, September 9, 2014 10:08:43 PM UTC-4, Boxiang Sun wrote: >>>>> >>>>> Hi all, >>>>> >>>>> Before the Google Summer of Code 2014, I proposed a proposal about >>>>> OpenCV-Julia binding. But at last, I was accepted by another project. Now >>>>> I >>>>> am successfully accomplished the GSoC. With the help of Tim Holy, I am >>>>> restarted the OpenCV-Julia binding. Tim just send me a link: >>>>> https://github.com/JuliaLang/julia/issues/8288. And I think I need to >>>>> let people know there has a project about OpenCV-Julia binding. >>>>> >>>>> During the time that I talk to Tim Holy, and my own research. I have >>>>> some results, please allow me to introduce the status of the binding. >>>>> >>>>> I am wrapped some basic API manually and tested it. Try to find a >>>>> correct mechanism of the binding. Now my plan is like OpenCV-Python >>>>> binding >>>>> did: >>>>> >>>>> *Extract OpenCV API, like the hdr_parser.py did in OpenCV-Python >>>>> binding. But rewrite the header parser with Julia(We don't want >>>>> OpenCV-Julia has dependency of Python, do we?). Another reason is the >>>>> output of hdr_parser.py is not perfect for Julia wrapper. >>>>> *Write automatic generation tool to wrap OpenCV API base on the output >>>>> of header parser. Generate the wrapped API, the wrapped API could build a >>>>> shared lib, just like OpenCV-Python did. >>>>> >>>>> I will continue try to wrap more API manually today. Then try to write >>>>> the header parser in Julia. That is the recently plan. >>>>> >>>>> Some details, such as memory management, interactive between >>>>> OpenCV-Julia binding and Image/Array, and other things, not decide yet. >>>>> >>>>> Any comments or suggestions will be highly appreciated! >>>>> >>>>> Regards, >>>>> Sun >>>>> >>>>
