Made with openFrameworks

I wanted to share this with you, is not a super cool graphics app… but is the proof that you can use oF to make all kind of things.

Is made in oF and the neural network part in openVino

Sorry is in spanish

https://vimeo.com/441003963

5 Likes

Hi Pandereto, that’s a pretty cool project.

I was wondering if you had any issues integrating openVINO with openFrameworks. I’ve made a couple of projects combining both tools (I’ve even made an openVINO addon for openFrameworks) but some of them don’t work properly when I try to run them on a Raspberry Pi with the Intel Neural Computer Stick 2. They run nicely in ubuntu with the NCS2, but in the raspi they crash when I try to perform a second inference.

Have you ever faced this kind of issue?

Best regards,
David

Hi David

Well openvino is a bit… tricky (dont know the exact english word) and has a lots of bugs i spend looots of time with that. Its super cool when it works :slight_smile: but when not is @#@#~@~

The issue exact as you describe i did not see it, but i can help if you want.

What openvino version are you using and what kind of model? Describe a bit your use or if you prefer send me a privete msg

1 Like

Thank you very much for your reply.

My project is a bit complex because I’m using a custom model and it needs some specific audio hardware to generate the input; there is more information in the thread I opened in the Intel forums. I am still working in an easier way to reproduce the issue, but I have a lot of things to do at the moment and I can’t spend too much time in this.

I assume that when you say in ubuntu works you mean it works with MYRIAD also, and you think it works until you look at the cpu usage doing inference on myriad…take a look and you will see a abnormal high cpu… is a bug. In raspbian it cost you one core 100%.

I have a openvino version compiled that can share in private as i dont know if i can share… it works perfect on x64 linux. It worked for me on a amr64 board i was going to try on raspbian os64 but run out of time so went to a x64 cheap celeron rugged nuc.

Do a fast test, export your model in irv7, download version 2019 R3 and try, that version is very good, later they incorporate ngraph or something and think began to get broken… There is a lot of people still running that version specially on rpi.