[Feature] Cross-platform deployment capability

[leemayi]

What's the feature?

1）start a train on a linux machine 2）export the onnx model on a linux machine 3）infer on a window machine. you can export the trt model directly without installing the python environment. Just like paddlepaddle‘s deployment

Any other context?

No response

[BIGWangYuDong]

We give some documents for users, such as https://github.com/open-mmlab/mmdetection/blob/3.x/docs/en/user_guides/deploy.md

[leemayi]

We give some documents for users, such as https://github.com/open-mmlab/mmdetection/blob/3.x/docs/en/user_guides/deploy.md

I know that, but we must install the python environment on Windows and export the model. I don't think this way is very convenient

[hhaAndroid]

@leemayi Deployment support is currently implemented via mmdeploy, you can ask in the mmdeploy repo