Configuration YOLO2 (ubuntu16.04+cuda8.0+opencv3.1.0)

Requirements have been installed Cuda 8.0 and OpenCV3.1.0

YOLO Official website

Configure Darknet

git clone https://github.com/pjreddie/darknet  cd darknet   make

If there is no error input

./darknet  

Get output

./darknet <function>  

Description Darknet Configuration Succeeded

Open the Makefile file and change the beginning lines to

gpu=1CUDNN=1OPENCV=1

Then query your GPU's computing power to see if Makefile is included, as follows my GPU computing power is 6.1, adjusted to:

Arch=-gencode arch=compute_30,code=sm_30       -gencode arch=compute_35,code=sm_35       -gencode Arch=compute_50,code=[sm_50,compute_50]       -gencode arch=compute_61,code=[sm_61,compute_61]       -gencode arch=compute_52,code=[sm_52,compute_52]

After recompiling, you can implement CUDA-and OpenCV-based compilation

download pre-training files

wget https://pjreddie.com/media/files/yolo.weights  

Test

./darknet Detect Cfg/yolo.cfg yolo.weights data/dog.jpg  

The result I get is:

Layer Filters Size Input Output0Conv +  3X3/1   608X608X3-608X608X +    1Max2X2/2   608X608X +-304X304X +    2Conv -  3X3/1   304X304X +-304X304X -    3Max2X2/2   304X304X -- theX theX -    4Conv -  3X3/1    theX theX -- theX theX -    5Conv -  1X1/1    theX theX -- theX theX -    6Conv -  3X3/1    theX theX -- theX theX -    7Max2X2/2    theX theX -- theX theX -    8Conv the  3X3/1     theX theX -- theX theX the    9Conv -  1X1/1     theX theX the- theX theX -   TenConv the  3X3/1     theX theX -- theX theX the    OneMax2X2/2     theX theX the- -X -X the    AConv +  3X3/1     -X -X the- -X -X +    -Conv the  1X1/1     -X -X +- -X -X the    -Conv +  3X3/1     -X -X the- -X -X +    theConv the  1X1/1     -X -X +- -X -X the    -Conv +  3X3/1     -X -X the- -X -X +    -Max2X2/2     -X -X +- +X +X +    -Conv1024x768  3X3/1     +X +X +- +X +x1024 +Conv +  1X1/1     +X +x1024 +X +X +    -Conv1024x768  3X3/1     +X +X +- +X +x1024 +Conv +  1X1/1     +X +x1024 +X +X +    AConv1024x768  3X3/1     +X +X +- +X +x1024 atConv1024x768  3X3/1     +X +x1024 +X +x1024 -Conv1024x768  3X3/1     +X +x1024 +X +x1024 -Route -    -Conv -  1X1/1     -X -X +- -X -X -    -Reorg/2     -X -X -- +X +X the    -Route -  -    inConv1024x768  3X3/1     +X +x1280 +X +x1024 -Conv425  1X1/1     +X +x1024 +X +X425    todetectionmask_scale:usingdefault '1.000000'Loading Weights fromYolo.weights ... done!Data/dog.jpg:predictedinch 0.070790Seconds.dog: the%Car: -%Truck: $%Bicycle: -%Init do opengl support available

Configuration YOLO2 (ubuntu16.04+cuda8.0+opencv3.1.0)