[indent]Hello,[/indent]
I’m working on GPU porting for real-time image processing. I’m looking for free (or low cost) libraries using CUDA or that can be used with CUDA. So far, I know GpuCV (a GPU porting of OpenCV) and the libs of GPU4Vision (FlowLib, VMFLib, …).
I also wonder what the “MCL image processing library” mentioned in the 1.3.3 section (p8) of the NVIDIA CUDA C Programming Best Practices Guide (2.3) is.
Thanks in advance for any help.
I tried to use GpuCV, but the project seems to make no further progress since the end of last year and most of their filters are implemented in shader languages and not in CUDA. The filters they have implemented in their experimental version are using CUDA 1.1 and I haven’t manged to compile their code with VS2008. The developers of GpuCV are using VS2005 and so there are also some libraries/dependencies missing for VS2008.
GPU4Vision is in development and they have put a new version of FlowLib, that is now compatible with CUDA 2.2 and VS2008 on their homepage.
The “MCL image processing library” is also new to me.
Indeed, I’m very interested as well. I’ve looked around but haven’t found any mention of it, except here on the forums and in that guide document.
I know this is an old thread, but I just happened across this – apparently that was a stale reference in the Best Practices Guide. I’ve removed it for the next version of the BPG.
Pretty much everybody will want to use the Runtime API at this point, especially since the driver and runtime APIs are seamlessly interoperable since roughly CUDA 3.1 (meaning you can now selectively use only the handful of driver API functions you really need when you need them and get the convenience of the runtime API for everything else).
–Cliff