I use opencv_traincascade for training using following parameters
opencv_traincascade -data cascade_model -vec ..\create_template\pos_vec_file -bg ..\BG\bg.txt -featureType LBP -w 32 -h 32 -numPos 5000 -numNeg 10000
I tried to give more memory 1024 Mb (256 by default), but in this case i have memory allocation error. So the problem is that at the first training stages it was ok, but then on stage 12, 13 and so on it takes 8 hours per stage. During train process i have 10-15% processor activity. I have TBB binaries in opencv lib, but i don’t really sure that my PC uses it. How can i tell to the PC, that i want to consolidate all PC resources on training process?
Thanks.
Do you have IPP installed. And I think not all the processes in training are parallel. You need to write your own code make it totally parallel.