-
Notifications
You must be signed in to change notification settings - Fork 187
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Voxelized data is shifted upwards #49
Comments
Good catch. I already merged your changes and will regenerate the data within the next days. Luckily, the error did not invalidate the results from the metrics. I will write here, when I changed the data and everything is ready for download. Thanks again for finding this problem and notifying us about it. |
You're welcome. Thanks for the nice work with the dataset. |
Did you already find the time to fix the data on the server? I'm unsure whether to submit with the 'old/shifted' voxel data or the 'new/fixed' data. |
As always, different other things needed my attention. I hope to find soon time to look at the issue. Sorry for the delay :( |
Hi @risteon, I finally managed to take the time to regenerate the data. I visually verified that the data looks okay. I generate the data now with the correct counter. Hope that resolves the issue then. You can now download the new data at http://semantic-kitti.org/dataset.html#download. Bonus: I also now generate all scans for the training data which people now download. |
Hi @jbehley, |
Dear authors of SemanticKITTI,
I've made the following observation:
The voxelized annotated point clouds (input grid, target grid, ...) are shifted one voxel upward due to a bug in the voxelizer
saveVoxelGrid
routine. This gets apparent when plotting the input point cloud together with the voxelized data.See this pull request for the SemanticKITTI voxelizer:
jbehley/voxelizer#4
The text was updated successfully, but these errors were encountered: