Question about the this baseline

[zkyyyyyyy]

Hi, thanks to your great work.
I found the baseline code of this Challenge based on bevformer you provide does not have alignment operation between prev bev and current bev. In bevformer repository it is defined in "transformer.py". But in "transformerOcc.py" I do not find it. So can I ask why?

[sephyli]

Hello! The code in this repository builds upon last year's baseline, which deliberately omitted the temporal component of BEVFormer.
It appears that the unresolved issue stems from BEVFormer's shift calculations being based on a y-forward coordinate system, while for the occupancy dataset, we utilize the nuScenes ego coordinate system, which is x-forward. Previous contributors likely did not have the time to make the necessary adjustments and opted to remove this feature for convenience.

[sephyli]

We have a fix about the coordinates issue at LaneSegNet repo. You can check it out here.

[zkyyyyyyy]

We have a fix about the coordinates issue at LaneSegNet repo. You can check it out here.

I try to run this code but got no key named "can_bus" in the dict ['img_metas']. Then I go to check the anno .pkl file, actually no 'can_bus' information in it. Can you help me with this issue?

[sephyli]

The OpenOcc dataset doesn't construct a can_bus tensor. You can canstruct one or modify the code as you like.

[sephyli]

Here is a doc for can bus.

[Icecream-blue-sky]

The OpenOcc dataset doesn't construct a can_bus tensor. You can canstruct one or modify the code as you like.

So will the current baseline(pre_bev is None by default) perform worse than the case using and aligning pre_bev ?