Support OpenVINO Toolkit 2021

[Luxonis-Brandon]

Start with the why:

Many bug fixes, feature improvements, etc. are present in the latest OpenVINO release (summarized here) which are beneficial to DepthAI users. Among these are capabilities to do audio, speech, etc. and a bunch of improvements to model conversion and support for newer frameworks such as TensorFlow 2.2.x

Move to the how:

We need to refactor our internal firmware to support the changes in OpenVINO Toolkit 2021. We will do so in our Gen2 system over the coming weeks, with a targeted release of support being December 2020.

Move to the what:

Support OpenVINO 2021, here.

[Luxonis-Brandon]

We have initial 2021 support in Gen1:

• 2021.1: https://github.com/luxonis/depthai/tree/openvino_2021.1
• Also 2020.4 is supported here: https://github.com/luxonis/depthai/tree/openvino_2020.4

[Luxonis-Brandon]

This is now fully supported in Gen2. And in fact many OpenVINO are supported, including 2020.1, 2020.2, 2020.3, 2020.4, and 2021.1.

See here for example usage:
https://github.com/luxonis/depthai-experiments/blob/master/gen2_examples/08_rgb_mobilenet.py#L18