The 2D detection in the first stage of this approach does not require retraining or fine-tuning, and any off-the-shelf method that can produce 2D boxes and radar point associations can be directly plugged in.