GSoC 2014 - OpenCV proposal by Wieland Morgenstern

[Wieland Morgenstern]

Dear Julia developers,

I'm a 26 year old Master's student at TU Ilmenau (Germany) studying Electrical Engineering and Computer Science, with a major in Robotics. You may not know me yet, although some of you will have read my proposal for Google Summer of Code, where I propose to bring an OpenCV binding to Julia:

http://www.google-melange.com/gsoc/proposal/review/student/google/gsoc2014/wm/5629499534213120

I'd love to bring over OpenCV to Julia, to have a start to get people hacking great CV algorithms and applications. I've given it quite some thought and have laid out in detail how I would approach the problem in the proposal, but I'm not fixed to do the OpenCV idea; I'm open to work in any area that overlaps with my studies and interests, that is: imaging, machine learning, graphics, high parallelism or data analysis. Also, I won't shy away from documentation work, as I think good documentation is of vital importance to get people involved and using the software.

- Yeah, but can you actually code? - 

I've interned as a Java developer and for nine months as an Objective C/C++ developer. For my Bachelor's Thesis, I implemented an object detection algorithm (HOG) for the GPU with OpenCL. I'm currently working as a graduate assistant at a Fraunhofer institute, programming embedded devices with C/C++ Code in the area of robotics/autonomous vehicles.

For my coursework and personal work I implemented small projects in many and a quite diverse set of languages, including C, C++, Clojure, Java, JavaScript, MATLAB, Objective-C, Python and R, so I'm not a stranger to different programming paradigms or styles.

I have a Github profile there: https://github.com/w-m/

And very occasionally answer on StackOverflow here: http://stackoverflow.com/users/463796/w-m

If you'd like further references, please contact me directly and I'll refer you to colleagues.

- Great, but are you the right fit for Julia? - 

I'm honest about not having any proven track record in Julia. But I'd really like to get involved, and I think that this is the purpose of GSoC. Putting out personal for-fun projects is a lot of work and even more work to promote them properly. I've been looking for an open source project to contribute to regularly. Julia seems to be a great fit for my skill set and I could see myself contributing long-term.

What else would you like to know about me? I'll be answering emails or as w-m on IRC/Freenode.

I'd love to work with and learn from you, thank you for considering my proposal,

Wieland

[Ivar Nesje]

When I go to the google-melange link I only get:

You are not logged in as the user in the URL.

I think there are some setting you have to set to make the application publicly available.

Ivar

[Wieland Morgenstern]

I can't find any switch/setting on melange to make the proposal public, so here's a copy of the text for now:

http://wieland.morgenst.de/proposal.html

-Wieland

[Maik Hoepfel]

Hi,

I have known Wieland for over 10 years now, and can only wholeheartedly recommend him both as a developer and as a professional to deal with. I know he was rarely able to open-source his work, but have often spoken to him about my work as maintainer of Python's most popular ecommerce framework (github.com/tangentlabs/django-oscar) and am confident that he'd be a good contributor.

Regards,

Maik Hoepfel

[Kevin Squire]

While I'm not part of the decision making process, my understanding is that prior involvement with Julia is a strong factor in making a decision (looking at coding ability, and how well one interacts with others). 

Why don't you look for an up-for-grabs issue in the Julia GitHub repo, or perhaps something in Images.jl, and implement it. 

Cheers, Kevin

[Wieland Morgenstern]

Hi Kevin,

thanks for the suggestion - of course I have considered that, but unfortunately it's not possible. Right now I need every hour of the day to finish up my Master's thesis. 

Creating a patch or a proper feature would take at least 1-2 days of orientation, planning, coding and testing to do it right, time I just can't spare. Half-assing something would neither help the project nor would it boost my proposal.

I realize it may sound fishy that I'm not willing to contribute right now, but then again it's not that I'm hiding something, you can browse all my online non-Julia projects and contributions.

If my lack of prior involvement with the project is the decisive factor of rejecting my proposal I'm unfortunately going to have to live with that, since I just couldn't live with dropping the ball on my thesis over trying to get a foot in the door with this project instead. It's somewhat ironic that while I would have a lot of time for the project itself (May-Aug) I can't put any more time in the application since it's the busiest period in my life as of yet.

Cheers,

Wieland

[Kevin Squire]

Hi Wirland,

That's perfectly understandable-your thesis definitely needs to come first. I was suggesting it mainly because there seems to have been a bit of interest in Julia, including others who have expressed interest in OpenCV, and this is the sort of (good) advice that has been given to many others. 

At any rate, best of luck on your thesis and your application!

Cheers, Kevin 

[svakSha]

You could also apply for esa-socis: http://sophia.estec.esa.int/socis/
HTH,
॥ svaksha ॥ http://svaksha.com