feat/wod dataparser #33
Conversation
Feat/AE-70 neurad wod * fix: output_folder from wod dataparser same as data path * Merge branch 'doc/git-hub-readme' into feat/AE-70-neurad-wod * clean: typos * clean: version of dataset used. * doc: add image resolution info for metrics * Merge remote-tracking branch 'origin/feat/AE-70-neurad-wod' into doc/git-hub-readme * Merged in doc/git-hub-readme (pull request georghess#3) doc: add image resolution info for metrics * update: merged main branch from github back prior pull request * doc: commented torch dynamo suppress error message * revert: reverted changes to neurad config steps * revert: reverted changes to neurad config steps on viewer side * fix: removed wod api from requirements and put back to readme because of too strict requirements * feat: added support for waymo-open-dataset * revert: change to .vscode/launch.json * revert: change to settings.json * fix: typo same variable two times in function signature * fix tiny caracter * Merge branch 'wod-dataparser' into feat/AE-70-neurad-wod Merge back minor changes from github * revert: vscode/*.json files * fix: typo in render.py * feature: added support for waymo-open-dataset * doc: added contributors to waymo readme * doc: explain waymo-open-dataset dependency installation * fix: added waymo-open-dataset package installation in Dockerfile * clean last changes from github Approved-by: Stanguennec, Julien
pyproject.toml
Outdated
There was a problem hiding this comment.
Hi, Thanks for providing the Waymo dataparser. I tried to build the code and found the pyproject.toml file has unsolvable conflicts when running pip install with python 3.10.
ERROR: Cannot install neurad-studio==0.1.0 and torchmetrics[image]==1.4.0.post0 because these package versions have conflicting dependencies.
The complete error message is:
INFO: pip is looking at multiple versions of torchmetrics[image] to determine which version is compatible with other requirements. This could take a while.
INFO: pip is still looking at multiple versions of torchmetrics[image] to determine which version is compatible with other requirements. This could take a while.
INFO: This is taking longer than usual. You might need to provide the dependency resolver with stricter constraints to reduce runtime. See https://pip.pypa.io/warnings/backtracking for guidance. If you want to abort this run, press Ctrl + C.
ERROR: Cannot install neurad-studio==0.1.0 and torchmetrics[image]==1.4.0.post0 because these package versions have conflicting dependencies.
The conflict is caused by:
neurad-studio 0.1.0 depends on torchmetrics>=1.0.1
torchmetrics[image] 1.4.0.post0 depends on torchmetrics 1.4.0.post0 (from https://artifactory.bluel3.com/artifactory/api/pypi/pypi/packages/packages/6d/e6/e51997d1818a4c1a1ad2b1c7ca5ff9dd95969596add58b2ed39479026964/torchmetrics-1.4.0.post0-py3-none-any.whl#sha256=ab234216598e3fbd8d62ee4541a0e74e7e8fc935d099683af5b8da50f745b3c8 (from https://artifactory.bluel3.com/artifactory/api/pypi/pypi/simple/torchmetrics/) (requires-python:>=3.8))
To fix this you could try to:
1. loosen the range of package versions you've specified
2. remove package versions to allow pip attempt to solve the dependency conflict
ERROR: ResolutionImpossible: for help visit https://pip.pypa.io/en/latest/topics/dependency-resolution/#dealing-with-dependency-conflicts
There was a problem hiding this comment.
Hi @kungfrank ,
I am surprised to see your error message, as neurad-studio does not require very tight version of torchmetrics[image], here: https://github.com/georghess/neurad-studio/blob/main/pyproject.toml#L56C1-L58C34, and in your error message.
Have you followed the installation guidelines from scratch with a fresh conda installation ? https://github.com/JulienStanguennec-Leddartech/neurad-studio/blob/feat/wod-dataparser/README.md
It looks like your python environement might already have some package of torchmetrics[image] package installed.
There was a problem hiding this comment.
Hi @JulienStanguennec-Leddartech
Yes, I got this error by following the new readme and creating a fresh conda environment. I have also tried it on two different desktops multiple times, but I keep getting the same issues.
There was a problem hiding this comment.
I recreated another new Conda environment from scratch by following your instructions again, and I got the same error again. Pip install tried all the way from torchmetrics-1.3.2 to 1.0.1 and failed in the end. Could you double-check this?
There was a problem hiding this comment.
Will do
There was a problem hiding this comment.
Hi @kungfrank, for your info, I have also tried to creat an env neurad by following the README in main branch, and got the same issue as well:
and followed your post, it now works:
we'll add this fix in the PR.
Thanks!
There was a problem hiding this comment.
I installed a clean env about a week ago and did not have any issues then (torchmetrics==1.4.0.post0) so I'll do some digging what has changed. Would be nice to keep down the number of steps in the installation
There was a problem hiding this comment.
I've pinned the issue down to the release of numpy 2.0. 96de2e2 should fix this. Could you check if you still have installation issues?
There was a problem hiding this comment.
I'll test this.
There was a problem hiding this comment.
@georghess, sorry for the late reply.
I've merge back the latest changes from your main branch, and tested installation and it seems to fix previously reported installation issues !
fix: code reviewed for PR in NeuRAD GitHub * fix: code reviewed for PR in NeuRAD GitHub Approved-by: Stanguennec, Julien
Pr reviewed Approved-by: Stanguennec, Julien
merge back changes from official repo
… repository changes
… with new requirements
|
@georghess the github workflow previously failed because of python version compatibility. I've updated the version python in the github/workflow, hopefully the CI/CD pipeline will pass. |
…package after waymo-open-dataset installed (req: numpy:1.21.5)
|
@georghess we are almost there ! ... sorry about the delay for doing the required changes. |
|
@georghess All checks have passed ! |
|
Thanks again @JulienStanguennec-Leddartech for providing this :) |
Hi ! Thanks for this great work.
We propose to add :
Readme with results and videos: nerfstudio/data/dataparsers/waymo_dataparser.md