Predict head pose from 2d106det.onnx ?

Hi, how can I predict head pose (roll, yaw, pitch) from landmarks predicted by `2d106det.onnx` ?

I see in the [code](https://github.com/deepinsight/insightface/blob/master/python-package/insightface/model_zoo/landmark.py#L66) that I need `1k3d68.onnx` model for that, but that model is 140 Mb while `2d106det.onnx` is only 5Mb.

I simply need to predict head pose fast and using small model