How to import custom models using eiq tool

Rabiraj — Wed, 11 Dec 2024 14:10:07 GMT

I have a face detect landmark based model, which works on my PC.
I would like to import into the mCXN947 MCU.
My model is based on Nonquantized ONNX model.

I tried to quantize it and convert it into a tflite-based model, then using xxd command, I tried to convert them into a C array and import.
I used label reader examples from eiq examples.

When I compile and load it, I am getting error :
AllocateTensors() failed!
Failed initializing model
Label image example using a TensorFlow Lite Micro model.
Detection threshold: 23%
Model: mobilenet_v1_0.25_128_quant_int8_npu
Didn't find op for builtin opcode 'PAD'
Failed to get registration from op code PAD

How to properly import a custom model and run it with a static images as input ?

Re: How to import custom models using eiq tool

Omar_Anguiano — Thu, 12 Dec 2024 22:06:54 GMT

Please refer to AN14241, there is detailed how to correctly import the model to EiQ so it can be used on MCX.

Best regards,
Omar

MCX MicrocontrollersのトピックHow to import custom models using eiq tool

How to import custom models using eiq tool

Re: How to import custom models using eiq tool