Ask Your Question

how to get cv.FaceRecognizer function for train() accelerate with CUDA?

asked 2017-02-08 02:27:41 -0500

faisal gravatar image

hai there, i would to ask about facial detection and recognizer in opencv 2.4. could it possible to accelerate it with CUDA on the function train() , because it takes a long time to train the datasets.

i am using python. i compiled my opencv 2.4 with CUDA 7.5, it is running fine.

i dont see any good python tutorial on opencv accelerated with CUDA? can u provide me the links ( if u have)

edit retag flag offensive close merge delete

1 answer

Sort by ยป oldest newest most voted

answered 2017-02-08 02:35:55 -0500

berak gravatar image

unfortunately, this might not be possible at all. even if you build your own cv2.pyd with CUDA, it won't be able to use any CUDA functionality from python (there is no "adapter" code for the CUDA classes)

then, the c++ face recognition code is not CUDA optimized, either (you'd have to completely re-write it)

edit flag offensive delete link more


sir, can i contact u somehow,i wanna ask u regarding accelerated opencv code. thx for answering my question

faisal gravatar imagefaisal ( 2017-02-08 02:38:14 -0500 )edit

so, the cuda is only optimized for C++ codes? not python? is it?

faisal gravatar imagefaisal ( 2017-02-08 02:41:50 -0500 )edit

yes, only for c++, and you have to write special code for it (using gpu_mat)

(also, rather ask your question here, so more ppl can help !)

berak gravatar imageberak ( 2017-02-08 03:00:43 -0500 )edit

ok2. thx for the information sir

faisal gravatar imagefaisal ( 2017-02-12 21:32:46 -0500 )edit

Question Tools

1 follower


Asked: 2017-02-08 02:27:41 -0500

Seen: 171 times

Last updated: Feb 08 '17