NPU Acceleration of Custom TFLite Model

I am attempting to run a simple compiled TFLite model on the Verdin IMX8mp, similar to the mobilenet demos in this link

I do not expect this model to run particularly well, as it is an unoptimized prototype. However, I am running into an _OpenDevice error as follows

root@verdin-imx8mp-scu-06965625:/bin# python3 decode.py
INFO: Created TensorFlow Lite delegate for NNAPI.
WARNING: Operator SQUARED_DIFFERENCE (v1) refused by NNAPI delegate: Unsupported operation type.
WARNING: Operator SQUARED_DIFFERENCE (v1) refused by NNAPI delegate: Unsupported operation type.
Applied NNAPI delegate.
[     1] Failed to open device: No such file or directory, Try again...
[     2] Failed to open device: No such file or directory, Try again...
[     3] Failed to open device: No such file or directory, Try again...
[     4] Failed to open device: No such file or directory, Try again...
[     5] _OpenDevice(1249): FATAL: Failed to open device, errno=No such file or directory.

The warnings are fine - as I still need to optimize the operations, but I’m not sure where to start with the “no such file or directory error”

Is this an issue with opening my TFLite model, the NNAPI backend, or what?

Beb

Hi @beb ,

Sorry for the delay to answer.

The No such file or directory is a bit strange, were you able to build your BSP with meta-ml and execute the demos provided on the layer?

Best regards,
Daniel Morais