To generate bin file without using edge impulse

Hey!

I’d want to know if I can generate a binary file for a trained model that can be used for openmv h7 module without using edge impulse. Any help would be appreciated.

TIA.

Hi, yes, you can. All cameras now support 100+ TensorFlow lite operators. You just need to generate an 8-bit signed quantized network.

That said, doing this requires understanding how to use the desktop tools to do so. Some of our users have done this via Google Colab.