Closed FiveLu closed 1 year ago
Thank you for bringing this issue to our attention. I apologize for any confusion caused by the lack of detailed instructions for dataset preparation in our README.
For a quick fix, I would suggest referring to the documentation in VoxFormer (where the preprocessed labels are provided, while the voxelized psuedo point cloud and query proposals are not necessary for our method) or MonoScene (you may need run the preprocessing yourself) for now. After that, don't forget to generate the depth prediction as referred to in our README.
As an additional note, I would like to mention that during our data preparation, we followed the instructions provided by MonoScene and have not specifically tested the VoxFormer dataset, but it seems they should be consistent.
We will work on improving the clarity and comprehensiveness of our README accordingly. Please let us know if you have any other questions.
I think I get it. I have run VoxFormer recently, so i could use
labels
andsequences
directly (modify the path in yaml). The extra thing I need to do is generate the folderdepth
. Thanks!
Hi. I am so confused about the structure of SemanticKitti dataset, especially
data_root
andlabel_path
. It's not elaborated in README. Could you tell me what it contains? And how to get it ?Thank you very much!