I got my hands on an Intel NCS (Neural Compute Stick) some weeks back and had been experiencing issues while trying to install and run the SDK. I decided to do this over the weekend, assuming it might take an entire day to actually making it run. Luckily, I was done with the setup and successfully running an example after 5 hours and going through countless resources, only to find out there was even a tar file available! This blog tries to summarize all the different hardwares on which the NCSDK can be installed and also about the two approaches one can use for the same.
On a Laptop/Desktop with Ubuntu 16.04LTS
There are two approaches via which one can get started with installing and using the NCSD. Since I chose the latter approach and faced many issues along the way, which took me some hours to find and rectify, I’d be mentioning them in this blog, so that no one has to scavenge through the Internet.
Download and extract the tar file
Use Git & make:
Issue 1: make install doesn’t work
After cloning the ncsdk repository and opening it, I was supposed to run make install, and that was it. However apparently it wasn’t that easy. On the first go I faced the issue –
Even after searching for hours was I not able to find a solution for this, since apparently no one faced any issues at this part of the installation. After consulting a friend I found out it was actually an issue with my drivers that was causing this issue.
sudo apt-get install libprotobuf-dev libleveldb-dev libsnappy-dev libopencv-dev libboost-all-dev libhdf5-serial-dev \libgflags-dev libgoogle-glog-dev liblmdb-dev protobuf-compiler
Issue 2: make example doesn’t work
git clone http://github.com/Movidius/ncsdk && cd ncsdk && make install && make examples
On an UP2 Board (With 16.04 LTS, 17.10 and 18.04 LTS)
Pretty straight forward actually. Once you’re done setting up the UP2 board, use the command mentioned above to download all the necessary libraries and examples.
On a RPi (Raspberry Pi) 2 Board (COMING SOON)