how to make a classifier inference on PX2 with tensorRT_optimization


We trained a classifier using digits, however, it seems that tensorRT_optimization is not working as it does not recognize the output layer.

How it would be possible to infer a classifer on PX2 board?
Thanks for your help.


Hi sacha,
You can provide output layer as a coomand line option . Please refer to TensorRT Optimization tool section in Driveworks development guide.