OpenVINO™ on the PC - The Prelude

While getting my hands dirty with OpenVINO™ and UP2 on the Edge, I got a bit side tracked and ended up working with a Windows* 10 Variant of the same. I decided to test out the functionalities being offered and how much of it's capabilities (heterogeneous execution and Computer Vision functions) could be leveraged.

KEY CONSIDERATIONS

The first and the foremost thing was to get going with the installation of OpenVINO in a Windows Environment and run some samples to see if everything was working as intended.

The first use case identified was to ensure the person identified can be cross referenced from an existing photo.

WHY I CHOSE OpenVINO™

Continuing on what I mentioned above, the Open Visual Inference & Neural network Optimization (OpenVINO) is a toolkit for developing applications and solutions that emulate human vision. Since this aligns with what I'm hoping to achieve out of the project, this was the best tool to start with. Moreover the availability of pre-trained models which had been optimized for OpenVINO ensured I could complete my task quickly.

PREREQUISITES

While the installation of OpenVINO's core components isn't much of a hurdle, it's the dependency of an IDE (Microsoft Visual Studio* 2015/2017) and Software components (Build Tools for the corresponding Visual Studio Version, Python* 3.6.X and CMake 3.4+) that requires some time and around 4-6 GB of Memory. You can refer to the link to learn more about the same and installation tips for OpenVINO on Windows.

NOTE - For some reason while installing CMake and selecting the appropriate option, i.e. Add CMake to the System Path for all Users, Cmake was still not being recognized as an internal or external command while running the Demo. The workaround for this is to enter the following in the command prompt:

"set PATH=C:\Program Files\CMake\bin;%PATH%"

For those looking to replicate the same on a Linux* machine, I've already written about the same earlier in an article here, with a small hack for enabling some functionality on an Ubuntu* 18.04 LTS machine.

MOVING FORWARD

Within 5 minutes of the complete installation, I was able to successfully run the demos (demo_squeezenet_download_convert_run.bat and demo_security_barrier_camera.bat) to verify OpenVINO's functionality with respect to facial recognition, and will come back with a new blog detailing the linking process with a Database.

For more complete information about compiler optimizations, see our Optimization Notice.