Feeding video recordings to the ai chip in the camera

Aingu7ae

Hello!

I'd like to use video recordings to test/evaluate the performance of the OAK-D Pro and the loaded model. Is there a way to feed h.265/h.264 streams (like the ones created in the VideoEncoder examples) into to chip in the camera? And if so, what's a good approach?

Kind regards

erik

Hi Aingu7ae ,
For AI-only evaluation, it might be easiest to use OpenVINOs Inference Engine (example here).

You could also stream video to the device and run inference on frames using depthai, example here. Note that video needs to be decoded on the host (I think openCV's VideoCapture already handles that). THoughts?
Thanks, Erik

Aingu7ae

Thank you, erik! This looks pretty much what I was looking for, I will expore both paths!