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
@zhiltsov-max proposed the idea to use datumaro transformations in CVATProjectDataExtractor class to reduce the amount of duplicated code with CVATTaskDataExtractor. Rough class implementation can look like:
task_datasets= { task_id: make_task_dataset(task) fortask_id, taskintasks }
# append name mangling and other transforms likefortask_id, task_datasetintask_datasets.items():
task_dataset.transform('rename', ...)
dataset=Dataset.from_extractors(*task_datasets)
The custom transformer can be written for frame number merging and image names mangling
My actions before raising this issue
@zhiltsov-max proposed the idea to use datumaro transformations in
CVATProjectDataExtractor
class to reduce the amount of duplicated code withCVATTaskDataExtractor
. Rough class implementation can look like:The custom transformer can be written for frame number merging and image names mangling
Originally posted by @zhiltsov-max in #3502 (comment)
The text was updated successfully, but these errors were encountered: