TensorRT + Python on TX2?

gong.ming.wei · June 26, 2018, 2:26am

According to the [url]https://devtalk.nvidia.com/default/topic/1029109/jetson-tx2/wheel-file-for-tensorrt-installation-on-nvidia-jetson-tx2/post/5236544/#5236544[/url]

it seems that Python API is only available on x86 Linux platform per Feb 2018.

So any plan to enable it works on TX2 platform?

If it doesn’t support TX2 platform, then what special procedures for us to develop TensorRT application for TX2 platform in Python?

Thanks.

AastaLLL · June 26, 2018, 6:57am

Hi,

Python API is not available for Jetson platform and we cannot release our future plan here.

Here is a sample can give you some idea of your requirement:

Python for pre-process
C++ for TensorRT inference
Swig as interface between python and C++

Thanks.

jkjung13 · May 20, 2019, 10:05am

I’ve created a TensorRT GoogLeNet example, in which I used Cython to wrap C++ code so that I could do TensorRT inferencing using python directly. Hope it helps.

Running TensorRT Optimized GoogLeNet on Jetson Nano: https://jkjung-avt.github.io/tensorrt-googlenet/
jkjung-avt/tensorrt_demos: https://github.com/jkjung-avt/tensorrt_demos

The code was tested on Jetson Nano, but it should work on Jetson TX2 too.

kayccc · May 23, 2019, 3:43am

Thanks for always sharing the awesome article to the community! Good job!