Config and code modification for BEVFormer w/OSP

[DengpanFu]

Hi, Thanks for your good job! I am trying to reproduce the result(mIoU=41.21) of BEVFormer w/OSP, but can not find the correct config. In your README-Results, OSP and BEVFormer w/OSP share the same config(projects/configs/osp.py), could you please provide the detailed config for BEVFormer w/OSP(e.g., the config to freeze trained BEVFormer)? Besides, I have some other questions:

1. why self-attn osp.py is not needed in OSP?
2. there are 3 encoder layers, but each layer forward twice BEVFormerEncoder, why? Any explanation is appreciated.
3. i find a hard-code use_bev_volume=False, is this should be True for using OSP as a plugin to argument volume-based methods? and how to modify the code?

Looking forward to your reply, Thanks!

[shawnsya]

Hi! The config for BEVFormer can refer to the baseline provided by the CVPR2023 Challenge.

1. The input query only interacts with the images and does not require self-attention.
2. This part can refer to the oversample and GPCA mentioned in the paper.
3. Thank you for pointing that out. This hard-code was added in a previous experiment and will be removed.