Re: [Final] Proposal by Olzhas Zhumabek - Epipolar geometry

Subject: Re: [Final] Proposal by Olzhas Zhumabek - Epipolar geometry
From: stefan (stefan_at_[hidden])
Date: 2019-04-08 13:36:52

Hi Olzhas,

On 2019-04-08 8:41 a.m., Olzhas Zhumabek wrote:
> Hi,
>
> I've submitted the proposal, with important details baked in. Link
> <https://docs.google.com/document/d/1cfj3mycgW1ItVO1DpsZiQRpeisUM0fvE3HszOV1VS5M/edit?usp=sharing>
> for last version.

Thanks !

> I've also deepened my knowledge about feature descriptors, and those seem
> to be very important in computer vision. Should feature detectors and
> descriptors go to GIL core? I would imagine many image transformations
> would like to keep them or select/erase those they want.

That's an interesting question. We definitely need to have a close look 
at the Boost.GIL code structure, and do some refactoring. There are 
different reasons for wanting to keep code in an extension rather than 
core. Added dependencies is one reason (so Boost.GIL users not needing 
those extensions don't have to install those dependencies).

On the other hand, if code ends up being itself a prerequisite for other 
extensions, it might be a good indication that it should actually live 
in the core. For example, some rather fundamental algorithms such as 
convolution right now live in the "numeric" extension. Arguably, this 
should be moved into "core".

In any case, I would suggest to start a separate discussion, and look at 
this case-by-case.

Best,

Stefan

--
       ...ich hab' noch einen Koffer in Berlin...