According my observation, the TensorRT Optimizer Tool included in DriveWorks-1.2 which is latest version for PX2 does not support importing ONNX. So I just wonder if I use newer version TensorRT Optimizer Tool to generate bin file and use it on DriveWorks-1.2 SDK DNN module, will it be OK?
Model generated from different DW release is not supported
So is there any solution for PX2s and DNN module to apply the pytorch model?
The support for ONNX model was added later which is not available on last release of DRIVE PX2. Only caffe and UFF models are supported on DRIVE PX2.
Ohh…It’s a sad story. It seems that I need learn toughTensorFlow.
Please consider upgrading to DRIVE AGX platform to get latest DRIVE SW updates.