Building TensorRT in Docker Python 3.8 Jetson Xavier AGX (JetPack 4.6.1)

Robotics & Edge Computing Jetson Systems Jetson AGX Xavier
, , , ,
1

I’m working with a Jetson Xavier AGX running JetPack 4.6.1, which includes Python 3.6 by default.

I’ve developed an object detection model based on YOLO and SAHI, and I want to deploy it inside a Docker container that uses Python 3.8, along with the corresponding versions of PyTorch, torchvision, and TensorRT.

However, I’m encountering issues during the image build process specifically, the failure occurs when building or integrating TensorRT with Python 3.8 in the Docker environment.

Could you please advise on the correct approach to make TensorRT work with Python 3.8 inside a Docker container on JetPack 4.6.1? Are there compatibility issues I should be aware of, or a recommended way to install TensorRT for this configuration?

3

Hi,

You can try to recompile the TensorRT pybinding to see if it can work.

Thanks.