search

Nightmare-n / GD-MAE

GD-MAE: Generative Decoder for MAE Pre-training on LiDAR Point Clouds (CVPR 2023)
Apache License 2.0
114 stars 6 forks source link

Support voxel-based GD-MAE? #13

Closed synsin0 closed 1 year ago

synsin0 commented 1 year ago

Thanks for your great work. In the code I see only the pillar implementation(assert height dimension is 1 in spt_backbone.py). I wonder whether you may support voxel-based implementation. Does GD-MAE support voxel-based embedding? Why and why not?

Nightmare-n commented 1 year ago

The MAE-style pretraining could also be used for voxel-based embedding (as far as I know, there is not much difference between voxel-based and pillar-based models). Since the baseline are pillar-based models (e.g., SPT), we only do experiments for them.