You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I'm trying to figure out how datumaro works, and eventually how to integrate to our dvc workflow. But I'm starting with a simple task: convert voc annotations to cvat format.
I'm creating a datumaro dataset with: pythonw -m datumaro create -d data/datumaro_dataset
then adding the one source with: pythonw -m datumaro source import -f voc -n v1 -p data/datumaro_dataset/ dir data/Workers/version_1/Annotations
then from the data/datumaro_datasetdir: pythonw -m datumaro source export -n datumaro_dataset -d ../export -f cvat
but I'm getting a: KeyError: 'Key "datumaro_dataset" does not exist or a: FileExistsError: [Errno 17] File exists: '/Users/vera/repositories/workers-detector/data/export'
Questions:
What will be the proper way to do this conversion?
I'm interested on contributing to datumaro; what will be the best communication channel with developers? (no really lucky on gitter)
@verasativa, currently, CVAT format support is not implemented in Datumaro, we are going to do it soon. Once it is supported, to export a dataset you would do:
datum project import -d data/datumaro_dataset -f voc data/Workers/version_1/Annotations
datum project export -p data/datumaro_dataset -f cvat ../export
Your variant with sources is also OK, but it should look like this:
pythonw -m datumaro create -d data/datumaro_dataset
pythonw -m datumaro source import -f voc -n v1 -p data/datumaro_dataset/ dir data/Workers/version_1/Annotations
pythonw -m datumaro source export -n v1 -d ../export -f cvat # note v1 here as source name
KeyError is for non-existing project source which was named v1, not datumaro_dataset.
Regarding FileExistsError: currently we avoid overwriting existing dirs to prevent data loss. We're open for your suggestions on design.
I'm trying to figure out how datumaro works, and eventually how to integrate to our dvc workflow. But I'm starting with a simple task: convert voc annotations to cvat format.
I'm creating a datumaro dataset with:
pythonw -m datumaro create -d data/datumaro_dataset
then adding the one source with:
pythonw -m datumaro source import -f voc -n v1 -p data/datumaro_dataset/ dir data/Workers/version_1/Annotations
then from the
data/datumaro_dataset
dir:pythonw -m datumaro source export -n datumaro_dataset -d ../export -f cvat
but I'm getting a:
KeyError: 'Key "datumaro_dataset" does not exist
or a:FileExistsError: [Errno 17] File exists: '/Users/vera/repositories/workers-detector/data/export'
Questions:
What will be the proper way to do this conversion?
I'm interested on contributing to datumaro; what will be the best communication channel with developers? (no really lucky on gitter)
cc: @zhiltsov-max
The text was updated successfully, but these errors were encountered: