I used Pointpillars for 3D object detection on my own dataset (just using *.bin files and their *.txt labels) using Pytorch. The format of the trained model is *.pth. How can I convert *.pth file to *.onnx format? I know I should define a model and a dummy tensor/input in this regard. How can I define these parameters for pointpillars?

I appreciate any help in advance.

Thank you,

Abbas