ONNX quantised Model

[mahanad]

Hi,

How can I reduce the inference time for an onnx model..  It's taking roughly 6 seconds.. 

I tried to quantise the model using eiq toolkit but when I tried to load the model it's giving me the following error...

terminate called after throwing an instance of 'Ort::Exception'
  what(): Fatal error: QLinearAdd is not a registered function/op
Aborted

Thanks in advance...

[mahanad]

Can it be used with onnx? According to the documentation, it's only for tflite (I might be wrong)

[Zhiming_Liu]

You need use vx delegrate in your inference code.

Please see this guide :

https://www.nxp.com.cn/docs/en/user-guide/IMX-MACHINE-LEARNING-UG.pdf