My apologies - I'm a few days late on replying, but here's a rundown:
- Graph-cut implementations (Cython and GPU) (I had not planned on implementing alpha extension, I'll have to to my list of nice-to-haves) - Grow-cut implementations (Cython and GPU) - QuickShift implementation (GPU) - A feasibility study of GPU algorithms and their integration into skim age
My full proposal is at https://google-melange.appspot.com/gsoc/proposal/review/google/gsoc2013/dekl...
Mmm, I'm not aware of any patent issues!! Could you point in the direction where I can find out!? Skimage isn't really planning on going the GPU route right now, Stéfan told me they did have a GSOC student a couple years ago that worked on creating an easily-switchable GPU backend, but it didn't materialize for a number of reasons. One of the goals of this project is to see weather we should look into it again though...
That's great! I could do with a hand - which seg algorithms did you do?
On Saturday, June 29, 2013 5:00:28 PM UTC+2, amue...@ais.uni-bonn.de wrote:
Hey Marc. I just saw that you are working on graph cuts and segmentation for skimage. That is pretty awesome. I'm super busy at the moment so I only sort of follow skimage at the time and I can't see melange. Could you shortly give an idea of what the goals of your GSOC are? Are you also planning to implement alpha-expansion? And what are the thoughts about the patent issues?
Also, I seem to have completely missed the fact that skimage wants to do gpu now. Are there any pointers on how this is planned / what is already implemented?
I implemented some of the segmentation algorithms in skimage and I'd really like to know what is happening and if I could help in any way. I also did some energy minimization stuff, but was a bit bummed out by the patent issues.
Btw, the slic implementation is not identical with the reference implementation. If you want to work on segmentation, I think this should really be investigated ....