Tutorial on Onnx model modification / TensorRT plugin development


I have a ONNX model but the conversion with TensorRT bug at a specific layer.

I am on Jetson nano so I cannot have lastest version on TensorRT…

So I would like to modify this layer on my Onnx model and replace it by a personal layer I could code using python or C, and integrate it under TensorRT …

Is it something which is possible to do ?
Can you please give me link of tutorial explaining how to do that ?

Thanks you !

Hi @rdpdo2002 ,
I am checking the details and share the response soon.


Thanks you very much really appreciate :-)