search

wzzheng / TPVFormer

[CVPR 2023] An academic alternative to Tesla's occupancy network for autonomous driving.
https://wzzheng.net/TPVFormer/
Apache License 2.0
1.19k stars 107 forks source link

How to apply the BEVFormer to the 3D OCC task？ #36

Open qiy20 opened 1 year ago

qiy20 commented 1 year ago

Thanks for your great work ! I am particularly interested in details about how to apply the 2D BEVFormer to the 3D OCC task？

huang-yh commented 1 year ago

We simply made predictions on 2D BEV feature map and assigned predictions to voxels according to their xy coordinates. Though we did not go further than that, it should be better to apply some lightweight decoder on the 2D BEV feature map to diversify predictions along the z axis.