Hello Everyone
I am working on Endoscopy videos on which i have to neglect out some blurred frames in the videos. i am running my videos on MATLAB . For a size of 1 GB video MATLAB takes an approximate time of an hour to clear out the blurred frames and give high qualtiy video as output.
I switched over to CUDA for cuttting down the time to run videos. How should i approach for that ?
What should i do to to built my own CUDA progams or else modify the examples to run these videos cutting down the time.
Is this possible using CUDA ?
Please help me.
Thanks
Please don’t post same question on different branches of this forum. This won’t help you to get better advice but will certainly annoy forum visitors.
Hve you checked the Matlab plugin that CUDA has? Matlab plugin link
I have limited knowledge about what your aim as well as your process - Matlab. But if I understand right, you would be having some algorithm to decide whether a frame is of high quality or one that should be discarded. If you can port that step using the C extension that CUDA-enabled cards can run, you might be able to get it done (even w/o the matlab plugin)
Matlab would definitely be easier since you would get LOTS of functions ready to use. But in that case you will have to make sure that the Matlab plugin that CUDA provides helps you out. I have not used it at all and hence cannot make any comments about it.
Good luck!
It will be easier to answer if you describe here main points of your code. Are you using FFT? Or may be all your code based on a sparse matrices?