Community
cancel
Showing results for 
Search instead for 
Did you mean: 
Highlighted
Beginner
485 Views

Raspberry pi Zero with ncs2

I want to implement openvino on raspberry pi zero . I have NCS2 . Could any one help me? is it possible that i can run ncs2 on raspberry pi zero or i have to shift with Google AIY vision kit? Please Help me....

 

0 Kudos
15 Replies
Highlighted
94 Views

Hello Jishang, 

 

Thank you for contacting Intel technical support, I can see that you want to use OpenVINO™ toolkit on Raspberry Pi Zero. 

 

You need to be sure that your Raspberry Pi* board includes ARM* ARMv7-A CPU architecture, you can check this using the command "uname -m". Also, you need to have installed Raspbian Buster 32-bit or Raspbian Stretch 32-bit.

 

It is important to know that the OpenVINO™ toolkit was tested in a Raspberry Pi 3; you can try using your Raspberry Pi Zero but keep in mind it has not been tested. The Pi Zero has limited resources and it may or may not work.

 

You can click here to access the OpenVINO™ toolkit for Raspberry Pi documentation.

 

Regards, 

 

Javier A. 

Intel Customer Support Technician 

A Contingent Worker at Intel 

 

Highlighted
94 Views

Sir, just to clarify, this is the correct link.

Highlighted
Beginner
94 Views

Hello Javier.

 

Greetings

 

Thanks for the quick response.

 

I am using the raspberry pi zero with WiFi. And it has armv6l. now what what is the process for the armv6l ? i got the link for the arm64. But for the armv6l i am not able to found.

 

Also Myriad 2 i think supported with the raspberry pi zero in Google AIY vision kit. Now if that is compatible with the raspberry pi zero(armv6l) why the NCS2(Myriad X) is not compatible with the raspberry pi zero(armv6l)

 

check the links for the references

 

https://forums.intel.com/s/question/0D50P00004NM07TSAT/45-raspberry-pi-zero-w-hat-with-a-movidius-my...

https://www.blog.google/topics/machine-learning/introducing-aiy-vision-kit-make-devices-see/

 

Please help for me

 

Thanks,

 

From,

Jishang

 

0 Kudos
Highlighted
94 Views

Hello Jishang,  

 

Thank you for your response. 

 

The OpenVINO™ toolkit for Raspbian* has only been validated on the Raspberry Pi 3. We have some guides to build the open-sourced version of OpenVINO™ on single board computers, however, there we do not have a specific guide for the Armv6l on the Raspberry Pi zero. The process may be similar to Armv7 but I have not tried this myself. 

 

  

Regards,  

  

Javier A.  

Intel Customer Support Technician  

A Contingent Worker at Intel  

 

0 Kudos
Highlighted
94 Views

Hi JDesa7, 

 

We would like to follow up on your case regarding implement OpenVINO™ toolkit on Raspberry Pi zero. In our last interaction, we provided you a guide with the steps to build the open-sourced version of OpenVINO™ on single board computers. We would like to know if you’ve been able to try our suggestion. In case you did, we would highly appreciate if you would let us know the outcome of these. We will be waiting for your response. 

 

 

Regards, 

 

Javier A.   

Intel Customer Support Technician   

A Contingent Worker at Intel 

0 Kudos
Highlighted
Beginner
94 Views

Hi,

 

Thanks for the follow up. The link which you gave that works partial. upto dldt make it is working. when i am applying for the make install giving me an error of CMake Error at ngraph/src/ngraph/cmake_install.cmake:49 (file):

 file INSTALL cannot find

 "/home/pi/Desktop/github/dldt/inference-engine/build/VERSION".

 

Call Stack (most recent call first):

 ngraph/src/cmake_install.cmake:43 (include)

 ngraph/cmake_install.cmake:66 (include)

 cmake_install.cmake:42 (include)

 

 

Makefile:73: recipe for target 'install' failed

 

thanks

From,

Jishang

 

0 Kudos
Highlighted
94 Views

Hi JDesa7, 

 

We understand the outcome of your test, but we cannot assure the steps will work in your system based on the hardware you are using. As I mentioned before, the instructions have been validated on a Raspberry Pi 3 with armv7l processor. Building on the Raspberry Pi Zero will require you to fix any issues that arise during the build process. Please note, that this is currently not supported. If you chose to go this route, please share your results with the community.  

 

Regards, 

Javier A.  

Intel Customer Support Technician  

A Contingent Worker at Intel  

 

0 Kudos
Highlighted
Beginner
94 Views

Hi,

Greetings.

 

I can understand that this is not developed by the intel. i have to install. The problem which i get i told you. But if i am removing that command. The device is communicating with the NCS2. and getting the almost same result as raspberry pi. The object detection ssd is working. Also yolo -async is working. If you think that i have achieved something. i can update the steps /send you the binary or library. the Thing which i am struggling is that whatever changes i have to in the code i want to cmake the dldt and then making again for the same. then it will generate the binary. then i can use my code. So I need the help if Make install could work so that i dint require all time to remake it.

 

For that what error i am getting i have mentioned to you.

 

 Error at ngraph/src/ngraph/cmake_install.cmake:49 (file):

 file INSTALL cannot find

 "/home/pi/Desktop/github/dldt/inference-engine/build/VERSION".

 

Call Stack (most recent call first):

 ngraph/src/cmake_install.cmake:43 (include)

 ngraph/cmake_install.cmake:66 (include)

 cmake_install.cmake:42 (include)

 

 

Makefile:73: recipe for target 'install' failed

 

Thanks.

 

From,

Jishang

0 Kudos
Highlighted
94 Views

Hi JDesa7, 

 

I don't believe I'm following your issue. It sounds like you are trying to modify our sample c++ demos and compile again? Or are you modifying the Openvino source code? 

I'm glad you were able to build the OpenVINO toolkit from source on the Raspberry Pi Zero, feel free to share your steps and results with the community. 

 

Regards, 

Javier A.  

Intel Customer Support Technician  

A Contingent Worker at Intel  

 

0 Kudos
Highlighted
94 Views

Hi JDesa7, 

 

I hope you are doing well. Do you still facing the same issue? If so, could you please answer the questions on my previous response? 

  

Regards, 

  

Javier A. 

Intel Customer Support Technician 

A Contingent Worker at Intel 

 

0 Kudos
Highlighted
Beginner
94 Views

i am modifying the demo sample code and making the application. Yes the library build for the raspberry pi zero.

 

Thanks.

From,

Jishang

+91-9022257710

0 Kudos
Highlighted
94 Views

Hi JDesa7, 

 

Thank you for your response. 

 

The instructions below are for building the open model zoo demos on raspberry pi, you can build all or individual ones. Use the command "make help" to list all the available options. The following instructions were tested on Raspberry Pi 4. 

 

git clone https://github.com/opencv/open_model_zoo.git cd open_model_zoo/demos/ mkdir build && cd build cmake -DCMAKE_BUILD_TYPE=Release -DCMAKE_CXX_FLAGS="-march=armv7-a“ ../. make all #builds all demos make object_detection_demo_yolov3_async #builds individual by name make help #lists all aviable options

 

 

Regards, 

  

Javier A. 

Intel Customer Support Technician 

A Contingent Worker at Intel 

 

0 Kudos
Highlighted
Beginner
94 Views

Hi,

 

Greetings. We cant use for raspberry pi zero

cmake -DCMAKE_BUILD_TYPE=Release -DCMAKE_CXX_FLAGS="-march=armv7-a“ ../.

 

as the architecture is deferent

 

thanks

From,

Jishang

0 Kudos
Highlighted
Beginner
94 Views

Hi,

 

Greetings. We cant use for raspberry pi zero. but more than 1 model can not run for the zero. E.g. pedestrian example or cross road examples. only yolov3 async and ssd async examples are working

 

thanks

From,

Jishang

 

0 Kudos
Highlighted
94 Views

Hi JDesa7, 

 

As we previously mentioned, the OpenVINO toolkit for Raspbian OS has only been validated on the Raspberry Pi 3 and 4. You can continue to try using OpenVINO with the Raspberry Pi Zero, however, this is out of our support scope. 

 

Regards, 

  

Javier A. 

Intel Customer Support Technician 

A Contingent Worker at Intel 

 

0 Kudos