use rf-detr from c/c++

[nlpvan-paul]

Hi,
I am familiar and use open-cv lib which is c++ based.
Can you port the python code to use your models from c/c++ program?
Thanks,
Paul.

[isaacrob]

You should be able to export to ONNX and then run either the ONNX via C++ or further convert to TensorRT.

We're not going to port the general codebase to C++ since we heavily leverage PyTorch

[nlpvan-paul]

Hi Isaac,Thank you for the quick replay.I don't use python and maybe am asking basic questions. Is the python to cython conversion, where the python code is converted to a .so lib, going to work with rf-detr? Thanks,Paul. On Tuesday, January 27, 2026 at 11:25:56 AM PST, Isaac Robinson ***@***.***> wrote: isaacrob left a comment (roboflow/rf-detr#602) You should be able to export to ONNX and then run either the ONNX via C++ or further convert to TensorRT. We're not going to port the general codebase to C++ since we heavily leverage PyTorch — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you authored the thread.Message ID: ***@***.***>

[isaacrob]

It is unlikely to work I think, but not sure!

What are you trying to do?

[nlpvan-paul]

I will look then at what you recommended initially.I did read on the new vision model and like to integrate it with opencv lib, on video frames. Paul. On Tuesday, January 27, 2026 at 06:56:35 PM PST, Isaac Robinson ***@***.***> wrote: isaacrob left a comment (roboflow/rf-detr#602) It is unlikely to work I think, but not sure! What are you trying to do? — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you authored the thread.Message ID: ***@***.***>

[isaacrob]

Got it.

Something like this might be a good starting point, though I haven't looked deeply this specific article https://learnopencv.com/how-to-run-inference-using-tensorrt-c-api/