Closed Fightinging1997 closed 2 weeks ago
Because our lane head is bezier-based, it outputs 4 control points and transforms them into 11 lane points using a transformation matrix. Both training and testing are the same.
Thank you for your answer, but I still have some questions. Where does transforms refer to here?
Please refer to this line: https://github.com/wudongming97/TopoMLP/blob/master/projects/topomlp/models/heads/lane_head.py#L731
Thank you for your answer.
Hi, thanks for open-sourcing the code!
I don't quite understand the meaning of the following paragraph in the paper.
LaneDetector. The lane query number is set to NL=300, and the number of control points is 4. Duringtraining, the control points are transformed into 11 lane points forcal culating loss.
Does the above sentence mean training by 11 points and inference by 4 points? If that's what it means, how do you do it?