Closed RoyiAvital closed 8 years ago
See Benchmark at the end of the file: https://github.com/Khanattila/KNLMeansCL/blob/master/DOC.md CPU is much slower, ~5/6 FPS maybe.
Hi, Even with AVX on Haswell / Broadwell / Skylake?
I'm talking about 1080p frames.
Thank You.
Haswell / Broadwell / Skylake and 1080p? ~1 fps maybe. On CPU use a smaller 'd' and 'a' is recommended. However, the use of a GPU is always recommended.
@Khanattila, Is there a way to contact you on email?
Thank You.
What's the matter?
Hi, I know someone who wants to have your advising on NLM.
Thank You.
@Khanattila , Did you get my last message?
Thank You.
Tell him to write directly on github.
Hi, It would be hard, he doesn't have an account and it wouldn't be appropriate.
Please, it is something you'd be happy to hear about.
Thank You.
Hi,
By the way, does it work on Floating Point images (Each channel as 32 Bit Float)?
Thank You.
yes of course
That's interesting.
Could it be adopted into OpenCV which lacks support for Floating Point?
Some changes should be made to work with OpenCV. Also there are some licensing issues.
Is there an issue to run it for University Project?
You can do anything the GPL permits: http://choosealicense.com/licenses/gpl-3.0/.
EDIT. You can really do a lot of different things =)
Hello, How many Frames Per Second can it yield on modern Intel CPU (Intel OpenCL Driver)? How about on modern GPU (Not the high end models)?
Thank You.