innovatefpga / 2019-pr072 Goto Github PK

to run this project,you need install OpenVINO toolkit.

how to run this model ?

[1]install OpenVINO toolkit 
[2]put src folder in /opt/intel/openvino/inference_engine/samples
[3]run command :./build_sample.sh
if compiled OK,you'll find a file named myprojectback in
~/inference_engine_samples_build/intel64/Release
[4]you can cd into this file run command :
./myprojectback -i cam0 -m ~/YOLO/bdd100k/tiny_yolov3.xml -d CPU 
to see the result of this project.
if you wan't see the result of videl ,what't you should to do is change cam0 to your  video path.



how to run this model ?
[5]if you wan't use FPGA to accelerate the project ,you need use command :
export CL_CONTEXT_COMPILER_MODE_INTELFPGA=3

aocl program acl0 /opt/intel/openvino/bitstreams/a10_devkit_bitstreams/2019R1_A10DK_FP16_TinyYolo.aocx


cd ~/inference_engine_samples_build/intel64/Release/


./myprojectback -i cam0 -m ~/YOLO/bdd100k/tiny_yolov3.xml -d HETERO:FPGA,CPU

you can see the accerate of FPGA(Arria10).


[6]we also offer the python project of tiny-yolo.you can see it in python folder.