Jingwei Zhang

Can not reproduce the performance of maptr with bevpool

7

Hi，the training of maptr with bevpool has some errors that some parameters are not used to compute loss, such as position_encoding. Althrough I have fixed them, the performance of maptr...

How about not sharing point-level queries in the one2many assigner

1

@LegendBC @wondervictor @outsidercsy Thanks for your code! I'm curious that sharing point-level queries could cause interaction between one2one group and one2many group in the training, which is not the desired...

global_translate_noise in CenterForm is different from that in CenterPoint.

4

Interesting work! The translation of data aug in CenterForm is 0.5, https://github.com/TuSimple/centerformer/blob/master/configs/waymo/voxelnet/waymo_centerformer.py#L132, while the translation in CenterPoint is 0. Also, I noticed that you used the `np.random.uniform` rather than `np.random.normal`...