Skip to content

Latest commit

 

History

History
87 lines (53 loc) · 3.45 KB

README.md

File metadata and controls

87 lines (53 loc) · 3.45 KB

Intel® Movidius™ Neural Compute Stick (NCS)

This lab shows how the Intel® Distribution of OpenVINO™ toolkit provides hardware abstraction to run the sample object detection application which was built in previous modules on Intel® Movidius™ Neural Compute Stick.

Connect Intel® Movidius™ Neural Compute Stick to your development laptop


image of Movidius NCS to computer


System check

First make sure the USB rules are set up.

cat <<EOF > 97-myriad-usbboot.rules
SUBSYSTEM=="usb", ATTRS{idProduct}=="2150", ATTRS{idVendor}=="03e7", GROUP="users", MODE="0666", ENV{ID_MM_DEVICE_IGNORE}="1"
SUBSYSTEM=="usb", ATTRS{idProduct}=="2485", ATTRS{idVendor}=="03e7", GROUP="users", MODE="0666", ENV{ID_MM_DEVICE_IGNORE}="1"
SUBSYSTEM=="usb", ATTRS{idProduct}=="f63b", ATTRS{idVendor}=="03e7", GROUP="users", MODE="0666", ENV{ID_MM_DEVICE_IGNORE}="1"
EOF


sudo cp 97-myriad-usbboot.rules /etc/udev/rules.d/

sudo udevadm control --reload-rules

sudo udevadm trigger

Then check if the device is visible with lsusb.

lsusb

The output will be If using NCS1,

Bus 002 Device 001: ID 1d6b:0003 Linux Foundation 3.0 root hub
Bus 001 Device 015: ID 03e7:2150

If using NCS2,

Bus 002 Device 001: ID 1d6b:0003 Linux Foundation 3.0 root hub
Bus 001 Device 015: ID 03e7:2485

Here ID 03e7:2150 without a description string is the Movidius device.

Setup a short path for the workshop directory

export SV=/opt/intel/workshop/smart-video-workshop/
source /opt/intel/openvino/bin/setupvars.sh

Run the sample application on Intel® Movidius™ Neural Compute Stick (NCS)

Set target hardware as Intel® Movidius™ NCS with -d MYRIAD

cd $SV/object-detection/

 ./tutorial1 -i $SV/object-detection/Cars\ -\ 1900.mp4 -m $SV/object-detection/mobilenet-ssd/FP32/mobilenet-ssd.xml -d MYRIAD

The Model Optimizer by default generate FP32 IR files if the data type is not particularly specified. In the previous command, we provided FP32 quantized model to Intel® Movidius™ NCS/NCS2 for inference. Technically, the Intel® Movidius™ devices support FP16 quantization only. The movidius plugin converts the FP32 quantized model to FP16 during runtime and does through any error even if you provide FP32 quantized model.

Let's run the Model Optimizer to get IR files in FP16 format to infer the model on the Intel® Movidius™ NCS by setting the data_type flag to FP16.

cd $SV/object-detection/mobilenet-ssd
mkdir -p FP16

cd /opt/intel/openvino/deployment_tools/model_optimizer

python3 mo_caffe.py --input_model /opt/intel/openvino/deployment_tools/open_model_zoo/tools/downloader/public/mobilenet-ssd/mobilenet-ssd.caffemodel

-o $SV/object-detection/mobilenet-ssd/FP16 --scale 256 --mean_values [127,127,127] --data_type FP16

Check if the .xml and .bin files are created in folder $SV/object-detection/mobilenet-ssd/FP16.

 cd $SV/object-detection/mobilenet-ssd/FP16
 ls

Now run the example application with these new IR files.

 cd $SV/object-detection/
./tutorial1 -i $SV/object-detection/Cars\ -\ 1900.mp4 -m $SV/object-detection/mobilenet-ssd/FP16/mobilenet-ssd.xml -d MYRIAD