Kevin Chen

I would recommend using the following tools [trtexec](https://github.com/NVIDIA/TensorRT/tree/master/samples/opensource/trtexec) or [polygraphy](https://github.com/NVIDIA/TensorRT/tree/master/tools/Polygraphy) for finer control of using TRT. The python backend in this repo is not as robust as these tools.

Thanks @jovialio for helping to root cause the issue. Since TRT does not accept the DOUBLE type at the moment, using the sizes parameter for the resize/interpolate op is a...

The change should be done in the ONNX model export step. However, in the latest TensorRT release we support DOUBLE / FLOAT types for resize scales. You can try the...

Gridsample is on our roadmap and will be supported in the next major TensorRT release.

I'm not too familiar with Yocto, maybe a set up issue somewhere?

I'm not sure if the onnx_tensorrt backend can support multithreading out of the box like the way you are using it. You probably need to follow the TensorRT APIs as...

Looks to be an error with protobuf - what version of protobuf do you have? Can you try building in the TensorRT container? - https://catalog.ngc.nvidia.com/orgs/nvidia/containers/tensorrt

This is an error that occurs with older TensorRT versions. As @nvpohanh mentioned, can you try upgrading to TensorRT 8.4.1 and importing your model again?

What version of protobuf are you using? You may have to use a lower version.

What's your use case for using an older ONNX version? Since the current build relies on submodules, if you wish to use a lower version you may need to manually...