Hi Alexandre, I have yet to compile an exhaustive list of algorithms but I am going to mention two at the moment. We can compile a list later too:
1. Assisted background removal: The paper on this from Facebook is https://arxiv.org/pdf/1703.06870.pdf . The ability to identify different objects in an image including foreground objects can assist users in removing the background object. I am specifically interesting in the removal of hair (which poses to be a really big problem). I believe there are already a few trained models on this already available in python libraries (TensorFlow). 2. Deep Style transfer: The relevant paper is at https://arxiv.org/pdf/1508.06576.pdf . This can be used to create 'Prisma like' transfer of style from one painting to another. I believe we may also have pretrained models already available here but we could try training the network specific to our needs. I think the idea is to do something along the lines of what Adobe is doing with Photoshop or in the wider context for their entire product line. Although we lack the funding or infrastructure to be a serious competitor, there are several researchers in the image processing/Machine Learning community who've written amazing papers over the years and open sourced their datasets. We should still be able to get something respectable using their research as reference. If there is nobody on the Machine Learning side in GIMP, I suggest you let me read a few papers and make some implementations but specific to what the GIMP users want. I am not an ML guy but there are people in my community who are using ML for science and there is a possibility I will be able to talk to some of them (who do image processing, but with sunspots instead of pictures of puppies). ML is a strange and wonderful thing, and people have tried doing everything with it. Sometimes it works, sometimes it doesn't. But we have to know what the GIMP community wants, then I will try to be organize our collective thoughts in a more technical document. Thanks, Maitraya. On Thu, Jan 17, 2019 at 5:01 PM Alexandre Prokoudine via gimp-developer-list <gimp-developer-list@gnome.org> wrote: > On Thu, Jan 17, 2019 at 2:14 PM Maitraya Bhattacharyya wrote: > > > > Dear devs, > > > > I have recently joined the mailing list because I wanted to contribute my > > two pennies to GIMP development (since I use it for my work). I had a > look > > at the proposed plan for GIMP and wondered if people would be interested > in > > including some popular AI algorithms for several image processing tasks. > > > > I would be interested in writing implementations of some of these > > algorithms into gimp if someone can commit to writing a frontend/GUI for > it. > > > > It would be great if we can make a list of these algorithms to implement > > and rank them according to priority. > > > > As for my background, I am a theoretical physicist making simulations for > > HPCs (in C/C++) and interpreting their data (in Python). I have a > > reasonable workstation to train neural nets, if necessary. Be warned > that I > > have never written a GUI software in my life and I don't know the GIMP > > codebase at all. I envision these to be standalone scripts which can be > > called in from the GIMP interface. > > > > Please let me know what you think. > > Hello and welcome :) > > What algorithms in particular are you thinking of? > > Alex > _______________________________________________ > gimp-developer-list mailing list > List address: gimp-developer-list@gnome.org > List membership: > https://mail.gnome.org/mailman/listinfo/gimp-developer-list > List archives: https://mail.gnome.org/archives/gimp-developer-list > _______________________________________________ gimp-developer-list mailing list List address: gimp-developer-list@gnome.org List membership: https://mail.gnome.org/mailman/listinfo/gimp-developer-list List archives: https://mail.gnome.org/archives/gimp-developer-list
