Large Scale GRM status?
Dear all, I would like to know what is status of Large Scale Generic Region Merging project? Some users request this feature in OTB. I think that it could be a great feature indeed. In addition, a great work has been done in implementing it in modern C++, using threads, distributed computation, etc.
A few years ago a bunch of us have worked on LSOBIA. Anybody knows the status of this project? I think it should be doable to propose at least an upgrade of GRM to allow the segmentation of large images, at small cost. Even if the application is not fully stable, I think that we should at least propose it as beta in next releases. I believe that the implemented GRM used MPI. But it should be possible to put some MPI stubs or something to allow the compilation without MPI (it could be great that MPI is optional: one could use the GRM on big HPC machines of course, however that's not the case for all OTB users). This way we could ship LS GRM/relevant LSOBIA apps in OTB without adding extra dependencies.
What do you think?