Re: [Boost][GIL][GSOC 2021] Contribute to the library

From: Mateusz Loskot (mateusz_at_[hidden])
Date: 2021-03-28 19:17:38

On Sun, 28 Mar 2021 at 05:36, sayan chaudhuri via Boost
<boost_at_[hidden]> wrote:
>                         My name is Sayan Chaudhuri. I am a final
> year,undergraduate student,in the Computer Science and Engineering
> department of IEM,India . I am eager to participate in the GSOC 2021 and
> willing to contribute to the Boost GIL library.
>
>                          I have already gone through the wiki list of Boost
> GIL listing the various image processing and computer vision algorithms
> that have not been implemented yet. I have also gone through the
> Contributing documents for BOOST GIL.

A good start.

> While going through the Boost
> Mailing List Archives, I found that Igor Antropov had suggested to develop
> ORB descriptor during his GSOC tenure but it has not been implemented yet..
> I am therefore eager to develop the ORB descriptor for Boost GIL during
> GSOC 2021. ORB is useful in many situations like object recognition and
> structure for motion and also it is suitable for real time applications
> unlike SIFT and SURF.

You are free (and welcome) to propose a project that is not based on
the list of suggestions on the Wiki.
Have a look at this, feel free to add/pick algorithms of your interest
https://github.com/boostorg/gil/wiki/Image-Processing-Algorithms

> I also found out that while using Boost , everytime I
> need to see the output , I need to save the output in order to view
> it,thereby unnecessarily wasting memory when I do not wish to save the
> output.  So, the functionality of imshow() for OPENCV is missing for Boost
> GIL. I am also eager to work on that functionality as well during GSOC
> 2021.

The idea of minimal GUI/viewer hosted as part of GIL codebase
has been discussed several times - it is not a bad idea.
To me, the biggest issue is to implement it with zero or as little as
possible dependencies, without requiring a complex GUI framework.

It could be implemented practically dependency-less using native X11
or framebuffer (like fim, feh) for Unix and using Win32 API for Windows

By the way, if you are using Visual Studio, you may find this extension
useful, developed by our fellow in Boost, Adam Wulkiewicz:
https://github.com/awulkiew/graphical-debugging/

 > Would therefore anyone is interested in mentoring me?

This year, I'm not able to offer myself as a mentor.
I may be able to co-mentor though.

>                          As my programming competency test for Boost GIL, I
> have implemented the KMeans algorithm from scratch using Boost GIL and I
> have raised a pull request- https://github.com/boostorg/gil/pull/587

Thanks, it will be reviewed.

> I would kindly request if someone can kindly review it. Also, I would like to
> know whether there are any similar issues  that I can work on currently.

You will need to browse the issues and find yourself.

Best regards,
-- 
Mateusz Loskot, http://mateusz.loskot.net