How to run a trained model using NCS2?

[Rimsha_ch]

hi,

i want some help about running a trained model on [RPI+NCS2]. i am new to NCS2 so not sure about this stuff. i tried to understand the generation of .bin and .xml files from guide provided here https://docs.openvinotoolkit.org/latest/_docs_IE_DG_Deep_Learning_Inference_Engine_DevGuide.html but that's just flying over my mind and i get nothing from that. i also visited other sites but not getting it. i wanted to run this model https://github.com/PINTO0309/OpenVINO-YoloV3 on [PRi] , which file exactly i needed to convert into .bin and .xml and how?

if you can help me with command line arguments so i can get something and able to use NCS2.

Thanks

[Luis_at_Intel · Accepted solution]

Hi @Rimsha ch,

Thanks for contacting us. There are some webinars available that overviews the essentials of the OpenVINO™ toolkit, you can find those webinar series here. I hope these videos help you get a bit more understanding about what IR format is (.bin and .xml).

As far as your question goes in regards of PINTOS' repo, you will first need to install OpenVINO™ toolkit on your Raspberry Pi board (if that is the route you want to go). He provides some steps in the repo under "Environment construction procedure" section, Option #2, or you can take a look at OpenVINO™ documentation for Raspbian here.

Once you have installed the OpenVINO™ toolkit and verified the installation, you will need to clone PINTOS' repository using:

cd ~/ git clone https://github.com/PINTO0309/OpenVINO-YoloV3.git

Navigate to this directory:

~/OpenVINO-YoloV3/lrmodels/tiny-YoloV3/FP16

Run the download_tiny-yolov3lrFP16.sh script to download the .xml and .bin files, as far as I can tell there is no need to convert any files to IR format as the script downloads them for you.

./download_tiny-yolov3lrFP16.sh

Then you can proceed to run the python sample for the NCS device with:

python3 openvino_tiny-yolov3_MultiStick_test.py -numncs 1

Note: Make sure you have a USB camera and your NCS device plugged in, otherwise may throw errors.

Hope you have found this helpful, let us know if you have any additional questions.

Regards,

@Luis_at_Intel​ 

[Rimsha_ch]

That helped me!! thanks you very much