Hi,

I would like to ask whether RF-DETR supports directly loading TensorRT .engine weights for inference.

Currently, the official sample code only supports loading .pt weights for inference. However, I plan to deploy the model using TensorRT (and later integrate an ID tracking pipeline), so I would like to run inference using the converted .engine weights instead.

The documentation and examples do not seem to mention how to load .engine files. May I ask whether direct TensorRT engine inference is currently supported, or if there is a recommended way to achieve this?

Thanks!