is it possibele to use 2 darknets one one machine with two Gpu ,s

asked 2020-06-15 09:52:12 -0500

updated 2020-06-15 10:20:50 -0500

HYPEREGO gravatar image

I want to use 2 seperate instances of darknet running on 2 different gpu's with one opencv(4.2) progam in c++ is that possible , and how ? the ./darknet -i 0 and ./darknet -i 1 works (-i selects on wich gpu it runs) thanks in advance

edit retag flag offensive close merge delete

Comments

Yes it seems possible - first of all compile dnn module with cuda support: https://www.pyimagesearch.com/2020/02.... You can even find help for this in the user forum.

You then need to specify the device id (the -i parameter on your yolo script) https://github.com/opencv/opencv/pull... with cv::cuda::setDevice

The most difficult part is the configuration(compiling). If this works for you - please close question as answered.

Please think also about using yolo natively from c/c++/python

holger gravatar imageholger ( 2020-06-15 18:59:06 -0500 )edit