Add script to generate normals GT, tweak dataloader, correct intrinsics #31
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Also adjust normals GT paths to match generate_normals.py output. Normals GT is now separate for left and right cameras. Add option to use camera image path format used by the `raw_data_downloader.sh` script.
valgur
force-pushed
the
feature/generate_normals.py
branch
from
9857f3a
to
8789cc7
Compare
|
The valgur/surface-normal fork now also includes CUDA support, which cuts down processing time significantly. Processing the full KITTI depth dataset now takes just 9 minutes instead of 2.5 hours on my laptop after this addition. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Thank you for open sourcing the model! It's much appreciated.
This PR adds a script to generate the surface normals ground truth required for training, which is currently not provided out of the box in the repo. It uses a custom version of the
surface-normalstool found here for this, which adds an easily installable Python wrapper for the code.I adjused the normals generation script to produce separate normals image for both of the left and right cameras and modified the dataloader scripts to match that (along with some other small tweaks).
I also corrected the camera intrinsic values to match the ones provided with KITTI (as mentioned in issue #30).