I used dusty’s jetson inference example to create a TensorRT Plan from FCN-Alexnet-Cityscapes-HD caffe-network.
After that I tried to use it in the DriveWorks framework by:
dwDNN_initializeTensorRTFromFile(&m_networkHandle, m_context, fileName);
I ran the example on my Computer (not on the DrivePX) with the Gencode:
Trying to read the TensorRT Plan on my computer I received this error:
TensorRT model file has wrong magic number. Please ensure that the model has been created by TensorRT_optimization tool in DriveWorks.
Of course this plan is not optimized for the DrivePX, but in my opinion I should be able to run it on my computer and load it into DriveWorks.
I have no idea why I get this error. I didn’t changed anything in the jetson inference example, so the plan should be correct.
The context is also working fine, I was able to initialize a bunch of other stuff.
Does anyone have an idea?
Thanks for your help and time!